Use of Non-Benzodiazepine Sedative Hypnotics and Risk of Falls in Older Men

Background: To ascertain whether use of non-benzodiazepine sedative-hypnotics is associated with risk of falls and compare this to risk of falls associated with use of benzodiazepines. Methods: Among 4450 community-dwelling men, aged 71 years and older, enrolled in the population-based prospective cohort study, Osteoporotic Fractures in Men (MrOS), use of non-benzodiazepine sedative-hypnotics and benzodiazepines was assessed by interview and verified from medication containers at the third annual visit of the MrOS study. Falls in the subsequent one-year period were ascertained by tri-annual questionnaires and a computerized dictionary used to categorize type of medication. Results: In age-adjusted models, non-benzodiazepine sedative hypnotic use was associated with an increased risk of any falls (one or more falls) (RR 1.44, 95% CI 1.15, 1.81) and recurrent falls (2 or more falls) (RR 1.51, 95% CI 1.07, 2.14). Use of benzodiazepines was associated with a similar increase in age-adjusted risk of falling. Depressive symptoms, inability to stand from a chair, and instrumental activities of daily living (IADL) impairment modestly attenuated these associations. The association between non-benzodiazepine sedative-hypnotic use and falls was most pronounced among men without a history of falls in the previous year: in a multivariable model controlling for multiple potential confounders, the RR of any falls was 1.74 (95% CI 1.13, 2.68) in this subgroup. Conclusions: Use of non-benzodiazepine sedative-hypnotics is associated with an increased risk of falls. Nonpharmacologic approaches to sleep disturbances may represent the safest approach to sleep difficulties in older adults.


Introduction
Nonbenzodiazepine sedative-hypnotics, such as zolpidem, zaleplon, and eszopiclone, are often advocated as safer alternatives to benzodiazepines for the treatment of sleep disturbances due to their short half-life and preservation of normal sleep architecture [1][2][3][4][5]. However, limited data are available about their safety in older patients, in particular regarding postural instability, falls, and fractures.
These so-called "Z-drugs" affect the same receptor as benzodiazepines, suggesting that their risks may be similar. Clinical trials of zolpidem in healthy younger adults have demonstrated central nervous system side effects, including impaired cognitive and motor function, particularly in the first few hours after use [6][7][8]. In addition, observational data have suggested an association between nonbenzodiazepine sedative-hypnotics and risk of fracture [9,10]. However, these studies have relied on administrative data bases and thus had limited or no information on important potential confounders, such as baseline physical function, cognitive function, and comorbidities.
To determine whether use of non-benzodiazepine sedativehypnotics is associated with an increased risk of falls, and to compare this to the risk of falls observed with benzodiazepine use, we ascertained use of non-benzodiazepine sedative-hypnotics and benzodiazepines in a cohort of 4450 men aged 71 years and older enrolled in the Osteoporotic Fractures in Men study (MrOS) and followed them prospectively for incident falls during one year of follow-up. Portland, OR; and San Diego, CA. Men with a history of bilateral hip replacement and men who were unable to walk without the assistance of another person were excluded [11,12].

Participants
To be included in the present analysis, men must have attended a 3rd clinic examination between March 2007 and March 2009, completed a medication inventory at the 3rd exam, and returned at least two follow-up questionnaires regarding falls in the subsequent one year period. Of the original cohort, 1043 men had died prior to the 3rd exam and 168 had terminated participation in the study; 4682 of the original cohort, (98% of survivors) attended the 3rd exam (baseline for this analysis). 4588 men returned at least two follow-up questionnaires in the subsequent one year period; of these, 4471 completed a medication inventory. 19 men who reported use of both a non-benzodiazepine sedative hypnotic and a benzodiazepine were excluded, as were two men reporting use of ramelteon, a melatonin receptor agonist. In a secondary analysis, we restricted the cohort to those 2722 men participating in an ancillary study evaluating sleep disorders, titled the MrOS Sleep Study, who had attended an earlier exam and completed a measure of subjective sleep quality (average 3.4 ± 0.5 years between ancillary sleep exam and 3rd exam).
The Institutional Review Board (IRB) at each center approved the study protocol and written informed consent was obtained from all subjects.

Medication use
Participants attending the clinic examination were asked to bring all current (any use within the last 30 days) prescription and nonprescription medications with them to clinic. Interviewers completed a medication history for each participant, including name of medication and frequency of use. All medications recorded by the clinics were stored in an electronic medications inventory database (San Francisco Coordinating Center, San Francisco, CA). Each medication was matched to its ingredient(s) based on the Iowa Drug Information Service (IDIS) Drug Vocabulary (College of Pharmacy, University of Iowa, Iowa City, IA) [13]. Zolpidem, zaleplon, and eszopiclone were categorized as non-benzodiazepine sedativehypnotics; benzodiazepines included lorazepam, clonazepam, alprazolam, temazepam, diazepam, triazolam, oxazepam, clorazepate, and chlordiazepoxide. Participants were categorized as users of a medication if they reported any use in the last 30 days.

Ascertainment of falls
At 4-month intervals, participants were queried by mailed questionnaire about the number of times they had fallen during the interval. Participants who fell in the previous 4 months were asked how many times (1, 2, 3, 4, or ≥5). Participants who did not initially return a tri-annual questionnaire or did not adequately complete the questionnaire were followed up with a telephone call. The present study includes fall reports during the one year period following the subject's Visit 3 examination.

Other Measurements
Participants completed questionnaires and were interviewed at the examination by trained staff, who asked about race/ethnicity, educational achievement, health status, smoking status, alcohol use, medical history, and falls in previous year. Men were asked whether they had difficulty performing any of 5 independent activities of daily living (preparing meals, shopping, heavy housework, walking 2-3 blocks, and climbing up 10 steps). A composite functional impairment score expressed the total number of activities ranging from 0 to 5 that a participant reported difficulty performing [14]. Physical performance measures included chair stand time and walking speed. For the chair stand, participants were asked to stand from a chair without using their arms; the time required to complete five chair stands without using arms was recorded [15].
Cognitive function was assessed with the Teng Modified Mini-Mental State Examination (mMMSE) (maximum score 100) [16]. Depressive symptoms were evaluated using the Geriatric Depression Scale short form (GDS) [17]. The total number of selected comorbid conditions reported by participants (stroke, diabetes mellitus, Parkinson's disease, chronic obstructive pulmonary disease, myocardial infarction, angina, and congestive heart failure) was summed for each participant.
Body weight was measured by using a balance beam scale at all sites, except at one site (Portland) which used a digital scale. Height was measured by using a wall-mounted Harpenden stadiometer (Holtain, UK). Body Mass Index (BMI) was calculated as weight in kilograms divided by the square of height in meters. Physical activity was assessed using the Physical Activity Scale for the Elderly (PASE) [18]. Among participants in the MrOS Sleep Study, sleep complaints were measured using the Pittsburgh Sleep Quality Index (PSQI), a 19-item self-report questionnaire that assesses sleep quality over a one month time period (range 0-21), with higher scores reflecting poorer sleep quality [19,20].

Statistical Analysis
Characteristics of the men at the 3 rd exam were compared by medication use category (non-user, benzodiazepine user, nonbenzodiazepine sedative-hypnotic user) using either analyses of variance or Kruskal-Wallis tests (continuous variables) or chi-squared tests of homogeneity (categorical variables).
For the primary analysis examining the association between medication use and risk of falls, log binomial and Poisson models were used to estimate the relative risk (RR) and 95% confidence interval (CI) of any falls (1 or more falls vs. 0) and recurrent falls (2 or more falls vs. 1 or none) by medication use category, with nonusers as the referent group. Models were initially adjusted for age and then further adjusted for potential confounders one at a time. In addition, covariates known to be risk factors for falls and characteristics related to use of sedative-hypnotics were examined for inclusion in a multivariable model, with those factors related to falls, nonbenzodiazepine sedative-hypnotic use, or benzodiazepine use at p <0.1 included. Interactions between medication and history of falls were explored. We also conducted secondary analyses in which the analysis was limited to the 2722 men who had data about self-reported sleep quality available from the earlier sleep exam.
impairments than non-users. They were more likely to have difficulty standing from a chair without using their arms and were more likely to have fallen in the previous year. Benzodiazepine users and nonbenzodiazepine sedative-hypnotic users were similar in many characteristics, although non-benzodiazepine sedative-hypnotic users were more likely to be college-educated, had higher mean MMSE, and reported more alcohol consumption.

Non-benzodiazepine sedative-hypnotic use and falls
In age-adjusted models, non-benzodiazepine sedative-hypnotic use was associated with an increased risk of any falls (RR 1.44, 95% CI 1.15, 1.81) and recurrent falls (RR 1.5l, 95% CI 1.07, 2.14). The inability to stand from a chair, and GDS all attenuated the risk of recurrent falls for non-benzodiazepine sedative-hypnotic users ( Figure  1a). Figure 1a-b: Association between Non-benzodiazepine Sedativehypnotic use, Benzodiazepine Use and Risk of Falls, model adjusted for age, site, GDS score, educational status, BMI, comorbidity index, ability to stand from a chair, PASE score, self-reported health status, IADL impairments, alcohol use, and MMSE.
In multivariable models that simultaneously adjusted for age, site, GDS score, educational status, BMI, comorbidity index, ability to stand from a chair, PASE score, self-reported health status, IADL impairments, alcohol use, and MMSE, use of non-benzodiazepine sedative-hypnotics appeared to be associated with modest increases in the risks of any fall and recurrent falls [(any falls: RR  There was a significant interaction between use of nonbenzodiazepine sedative-hypnotics and history of falls in the ageadjusted model for any falls (p=0.01); for recurrent falls the interaction was not significant (p=0.24), possibly due to limited power. We subsequently stratified on history of falls in the previous year. The risk of falling with non-benzodiazepine sedative hypnotic use appeared to be most pronounced among men without a falling history. For men without a history of previous falls, non-benzodiazepine sedative hypnotic use was associated with an increased risk of any falls in ageadjusted models (RR 1.83, 95% CI 1.23, 2.70) (Figure 2). The point estimate for the age-adjusted relative risk of recurrent falls (RR 1.77, 95% CI 0.83, 3.79) was similar to that for the risk of any falls but did not reach significance, likely due to low power. In the multivariable model, non-benzodiazepine sedative-hypnotic use was significantly associated with an increased risk of any falls [RR 1.74 (95% CI: 1.13, 2.68)]; for recurrent falls, the relative risk was 1.61 (95% CI: 0.62, 4.13). For the subset of men who had previously fallen, there was no evidence of an association between use of non-benzodiazepine sedative-hypnotics and risk of subsequent falls in age-adjusted or multivariable models. Figure 2: Association between Non-benzodiazepine Sedativehypnotic Use and Risk of Falls among men without a history of falls, model adjusted for age, site, GDS score, educational status, BMI, comorbidity index, ability to stand from a chair, PASE score, self-reported health status, IADL impairments, alcohol use, and MMSE.

Benzodiazepine use and falls
In age-adjusted models, benzodiazepine use was associated with an increased risk of any falls (RR 1.35; 95% CI 1.12, 1.61) and recurrent falls (RR 1.40; 95% CI 1.06, 1.85). The addition of IADL impairments, ability to stand from a chair, and GDS score each attenuated the association between benzodiazepine use and risk of any and recurrent falls (Figure 1b). Use of benzodiazepines was not associated with risk of any falls [RR 1.09 (95% CI: 0.90, 1.33)] or risk of recurrent falls [RR 1.08 (95% CI: 0.79, 1.47)] in the full multivariable model. Addition of the PSQI to the multivariable model in the subset of men with this measurement did not significantly alter the results. There was no evidence of an interaction between benzodiazepine use and history of falls (p=0.41 for any falls and p=0.70 for recurrent falls).

Discussion
In this cohort of older men, use of non-benzodiazepine sedativehypnotics and use of benzodiazepines were each associated with an increase in the age-adjusted risk of falling. The association between benzodiazepine use and falls appeared to be largely explained by greater disability, poorer physical performance and greater level of depressive symptoms among men taking benzodiazepines. In contrast, the association between non-benzodiazepine sedative-hypnotic use and falls was only modestly attenuated by consideration of potential confounders. Furthermore, this association depended on fall history and was most pronounced among men without a fall history. These findings are consistent with the limited existing epidemiologic data. In a case-control study of patients aged 65 years and older, investigators found a significantly increased risk of hip fracture in users of zolpidem (adjusted OR 1.95; CI 1.09-3.51) [9], a rate similar for benzodiazepines. A recent retrospective cohort study found that the risk of nonvertebral fractures and dislocations was higher in the 90 days following an initial prescription for zolpidem than the risk prior to the prescription in members 65 years of age and older; this increased risk for zolpidem was similar to or higher than that associated with several benzodiazepines [10].
Due to its observational nature, our analysis has several limitations, most notably confounding by indication. These medications are typically prescribed for sleep disturbances. Because sleep disturbances have been linked to an increased risk of falls [25], the observed association between use of non-benzodiazepine sedative-hypnotics and risk of falls may be due to the underlying sleep disturbances that prompted their initiation. To address this issue, we controlled for the PSQI in multivariable models and did not observe any significant change in our results. However, the PSQI was not available at Visit 3 in the cohort and thus we utilized a PSQI measurement from an earlier visit in a subset of our original analytic cohort.
Another potential source of confounding is the possibility of channeling bias -i.e., if these drugs were preferentially prescribed to patients with a higher risk of falls due to the perception that these drugs have a safer side effect profile than benzodiazepines. We attempted to address these issues by controlling for multiple potential confounders, as well as by excluding subjects who had reported falls in the year prior to the study visit. Nevertheless, the potential for unmeasured confounding remains a limitation of this analysis.
Our analysis is also limited by a lack of detailed information about the reported falls. Information regarding time of day of the falls or the mechanism of the falls is not available in the cohort. We also do not have information about dose of medication or frequency of use; it is possible than subjects were not taking these medications at the time of a fall.
We found a weaker association between benzodiazepine use and risk of falls than previously reported. This finding may be due to temporal trends in benzodiazepine prescription patterns, due to growing recognition of the adverse effects of benzodiazepines in the elderly. [26] We also found that the association between use of nonbenzodiazepine sedative hypnotics and risk of falls was more pronounced for men who had not fallen in the previous year. This finding may be due to the strong association between previous falls and risk of subsequent falls: the additional risk of a sedative hypnotic may be negligible for men already falling, whereas for men not previously falling, the addition of one of these medications may "tip the balance." Despite these limitations, this work adds to the growing body of literature suggesting that use of the non-benzodiazepine sedativehypnotics may be associated with adverse effects such as falls. Future randomized clinical trials of this class of medications should include falls as a safety outcome. In addition, non-pharmacologic approaches to sleep disturbances, such as cognitive behavioral therapy, may represent the safest approach to this common problem in older adult.