Sleep characteristics and use of multiple benzodiazepine receptor agonists in older adults

Long-term use of benzodiazepines or z-drugs (benzodiazepine receptor agonists; BZAs) is not advised for older adults due to adverse outcomes.1, 2 While studies of BZA prescribing patterns show increasing rates of co-prescribing of BZAs (e.g., benzodiazepines with z-drugs),3 to our knowledge, studies examining patient-reported rates of use of more than one BZA (multi-BZA) are lacking. This study aims to estimate rates of patient-reported multi-BZA use among older adults and to examine associations between multi-BZA use and sleep characteristics. Telephone screening surveys were conducted for an ongoing BZA deprescribing clinical trial (NCT03687086)4 to recruit adults aged ≥55 in California who were receiving care through the Department of Veterans Affairs or University of California, Los Angeles and were prescribed at least one of five targeted BZAs (alprazolam, clonazepam, lorazepam, temazepam, zolpidem) chosen for practicality for the dose reduction component of the trial. Screeners between February 2019 and October 2022 were analyzed. Our institutional review boards approved the study, including a waiver of documented consent. Participants reported whether they used each targeted BZA for sleep over the past 3 months (yes/no for each) and were classified as multi-BZA (>1), single-BZA (1), or non-BZA (0) users based on the number of “yes” responses. BZA use frequency was estimated by asking participants how many nights in a typical week during the past month they had taken each BZA. We did not assess for concurrent use in the multi-BZA group. Participants were asked if their sleep problems have lasted <3, 3–12, or >12 months (dichotomized as ≤3 months and >3 months) and if their sleep quality in the last month was “very good,” “fairly good,” “fairly bad,” or “very bad” (dichotomized as “very/fairly good” and “very/fairly bad”). Self-reported time asleep and time in bed were used to compute sleep efficiency (time asleep/time in bed), which was categorized as excellent (≥90%), good (≥85%), and fair (≥80%).5, 6 Analyses compared sleep characteristics between multi-, single-, and non-BZA users in unadjusted and adjusted (for site, age, gender, race, ethnicity) logistic and linear regression models (alpha = 0.05). Pairwise comparisons and predicted margins for comparisons (multi- versus single-BZA or non-BZA) were estimated and adjusted predicted margins from adjusted models plotted using Stata/SE 17.0 (StataCorp LLC, College Station, Texas). We screened 793 of 8514 potential participants (see Table 1 for sample characteristics). Multi-BZA was reported by 13.9% of participants in the past 3 months. Among multi-BZA participants, 87.3% reported taking 2 BZAs, most commonly alprazolam-zolpidem (28.1%) and lorazepam-zolpidem (25.0%). Most participants (96.5%) reported sleep problems greater than 3 months and fairly/very bad sleep (68.0%). Mean sleep efficiency was 67.7% (SD = 18.3%). Only 10.5% had high sleep efficiency (≥90%). In unadjusted models, multi- versus single-BZA status was associated with a 9.8% increase in poor sleep quality (p = 0.03). There were no other significant differences in sleep characteristics for multi- versus single- or non-BZA users (p-values ≥0.51). No differences were observed for multi- versus single- or non-BZA users for categorical sleep efficiency variables (p-values ≥0.07). In adjusted logistic and linear regression models, predictive margins for multi- versus single- or non-BZA were not significant for sleep problem chronicity, sleep quality, and sleep efficiency as a continuous variable (p-values ≥0.07). Figure 1 summarizes these findings. We found 1 in 8 older patients prescribed a BZA for sleep reported multi-BZA use over the past 3 months. Multi-BZA use was unrelated to chronicity of sleep problems, sleep quality, or sleep efficiency, suggesting that multi-BZA use was not associated with improved or worse sleep. The findings reflect patient reported use of BZA medications rather than prescriptions. Regardless of whether multi-BZA use was simultaneous or staggered, these findings are concerning since elimination half-life of these medications, which can be longer in older adults, may lead to overlap even when these medications are taken at different times or days. Amplification of adverse effects (e.g., cognitive impairment, falls2) is a concern during these periods of overlap. Through our experiences screening patients, we understand multiple factors might contribute to multi-BZA use. Some patients take benzodiazepines for anxiety and z-drugs for insomnia. Some patients whose medication quantity is limited request another medication from the same class to have “enough,” or patients may have multiple prescribing clinicians who are unaware of other BZA prescriptions. Study limitations include not recording reasons for multi-BZA use, inquiring about five BZAs only, not assessing concurrent use, and focusing on insomnia indications, thereby potentially underestimating multi-BZA use in daytime users. Future research should address these limitations to better characterize multi-BZA use. Sara Ghadimi: Conception and design, acquisition of data, analysis and interpretation of data, manuscript drafting, final approval of manuscript. Austin Grinberg: Acquisition of data, analysis and interpretation of data, manuscript drafting, final approval of manuscript. Michael N. Mitchell: Analysis and interpretation of data, manuscript drafting, final approval of manuscript. Cathy Alessi: Analysis and interpretation of data, manuscript drafting, final approval of manuscript. Alison A. Moore: Analysis and interpretation of data, manuscript drafting, final approval of manuscript. Jennifer L. Martin: Analysis and interpretation of data, manuscript drafting, final approval of manuscript. Joseph M. Dzierzewski: Analysis and interpretation of data, manuscript drafting, final approval of manuscript. Monica Kelly: Analysis and interpretation of data, manuscript drafting, final approval of manuscript. Safwan Badr: Analysis and interpretation of data, manuscript drafting, final approval of manuscript. Andrew Guzman: Analysis and interpretation of data, manuscript drafting, final approval of manuscript. Jason P. Smith: Analysis and interpretation of data, manuscript drafting, final approval of manuscript. Michelle Zeidler: Analysis and interpretation of data, manuscript drafting, final approval of manuscript. Constance H. Fung: Conception and design, acquisition of data, analysis and interpretation of data, manuscript drafting, final approval of manuscript. We are grateful for the contributions and efforts of the sleep research staff at VA Greater Los Angeles Geriatric, Research, Education and Clinical Center and UCLA. The authors declare no conflicts of interest. The National Institute on Aging and the Department of Veterans Affairs did not have any direct role in the analysis or writing of this manuscript. The findings do not necessarily reflect the views of the NIA or Department of Veterans Affairs. This work was supported by NIA R01AG057929 (to Constance H. Fung), VA HSR&IIR 17-234 (from the United States Department of Veterans Affairs to Constance H. Fung), UCLA CTSI UL1TR001881, VA Research Career Scientist Award (VA HSR&D RCS 20-191 from the United States Department of Veterans Affairs to Jennifer L. Martin) and NIH/NHLBI (K24 HL143055 to Jennifer L. Martin). The contents of this paper do not represent VA or NIH views or the United States Government.