No End in Sight: Benzodiazepine Use in Older Adults in the United States

Donovan T. Maust, MD, MS; Helen C. Kales, MD; Ilse R. Wiechers, MD, MPP, MHS; Frederic C. Blow, PhD; Mark Olfson, MD, MPH


J Am Geriatr Soc. 2016;64(12):2546-2553. 

In This Article


These analyses demonstrate that the population-based rate of new benzodiazepine use in U.S. office-based medical practice is generally consistent across adult age groups. Despite evidence of the harms associated with benzodiazepine use in older adults, efficacy of alternative pharmacotherapy and psychotherapy,[2–4,28] and professional guidelines advising against benzodiazepines,[11,12] initiation continues unabated, and continuation prescriptions account for a growing proportion of use with older age.

Prior work has suggested that benzodiazepine use in older adults is primarily for anxiety or insomnia.[15,16,22] The current findings confirm that anxiety and insomnia are the most common diagnoses and reasons reported for a benzodiazepine visit, at 21.3% and 11.6%, respectively, of new benzodiazepine visits, although there was no mental health–related diagnosis or reason for visit reported in more than 80% of all benzodiazepine visits and in nearly 60% of visits at which a new benzodiazepine was started. It may be that some benzodiazepines are started for brief stress or adjustment reactions that clinicians are not recording as a diagnosis. In a previous analysis of community-dwelling individuals newly started on an anxiolytic, the most common reason reported after anxiety or insomnia was "stressful life event, adjustment, [or] grief,"[16] although if the individuals reported a visit reason such as stress or tension, this should have been captured in the survey data (Table S2, code 1100.0). Another possibility is that older adults may report not anxiety but instead a variety of somatic symptoms.[29] Providers may conceptualize and treat such symptoms as anxiety but not label them as such. The continuation benzodiazepine user appears to be taking "benzodiazepine[s] for less specific indications" than well-defined psychiatric disorders, as noted previously.[22]

What is perhaps most notable—and has not previously been reported to the knowledge of the authors—is the extremely small proportion of individuals who received or were referred to psychotherapy. Cognitive behavioral therapy,[28,30] short-term psychodynamic therapy,[31] and other psychosocial treatments[29] are all effective for anxiety disorders, and cognitive behavioral therapy is effective for insomnia.[4] Despite the efficacy of psychotherapy to treat the most common conditions for which benzodiazepines are prescribed, fewer than 1% of overall and new benzodiazepine visits included provision of or referral to psychotherapy. Nonpsychiatrist physicians may benefit from additional education about the effectiveness of evidence-based psychotherapies and, perhaps more importantly, improved access for their patients to such treatments. NAMCS may underestimate the provision of psychotherapy, because people may be engaged in psychotherapy with another provider, or psychotherapy may have been previously discussed or attempted. In addition, nonphysician providers provide a significant amount of psychotherapy, which NAMCS does not capture. Nevertheless, adults aged 65 and older have the lowest rates of psychotherapy use of any age group, and the NAMCS-based rate is similar to a previous analysis of psychotherapy using data from the Medical Expenditure Panel Survey.[32]

In addition to psychotherapy as an alternative to benzodiazepines, selective-serotonin reuptake inhibitors are considered first-line pharmacotherapy for anxiety disorders.[33,34] Among all benzodiazepine visits, just one-quarter of individuals were taking an antidepressant. The low proportion of participants receiving psychotherapy and the low proportion of those taking antidepressants suggest that older adults are not receiving treatments that are more appropriate and safer than benzodiazepines.

Nearly 10% of those prescribed a benzodiazepine were also prescribed an opioid, and co-prescription of antipsychotics occurred in 3% of older adults. Co-prescribing of such central nervous system–active medications is associated with cognitive decline,[35] and use singly[5,6,36] or in combination[11] is associated with greater risk of falls and fractures. An added concern is their role in pharmaceutical overdose deaths. Individually, opioids and benzodiazepines are involved in more than 75% of pharmaceutical overdose deaths.[37] In opioid-related deaths, which are the leading type of pharmaceutical death, benzodiazepines are the most commonly co-prescribed medication, involved in 30% of deaths. Although combination use of these agents may be appropriate in select populations, use should be considered only after fully discussing potential risks, benefits, and alternatives.

There were few differences in other clinical characteristics between new and continuation benzodiazepine users. A larger proportion of continuation users have asthma or COPD, which is consistent with the high prevalence of anxiety disorders in individuals with chronic respiratory problems,[38] although it is also possible that respiratory symptoms may be misattributed to anxiety, leading to benzodiazepine treatment. Nonetheless, use of benzodiazepines in individuals with COPD is troubling given the association with mortality.[39]

Likewise, there are few visit characteristics that distinguish between benzodiazepine users and nonusers or new and continuation users. Visits with benzodiazepine users were slightly longer than those with nonusers, and new visits were slightly longer than continuation visits, but neither comparison was statistically significant or clinically meaningful. There was no difference according to visit disposition or whether the individual was established with the practice. Those taking a benzodiazepine had had more visits in the prior 12 months than nonusers, which is consistent with their higher overall medical comorbidity, but there was no difference between new and continuation benzodiazepine users.

This work has several limitations. First, individual-level clinical assessments of current symptoms and function were not available. Second, visit diagnoses were limited to three, so in individuals with more active problems, there may be information bias that affects the various benzodiazepine groups differently. However, when limiting analysis to encounters with no more than two diagnoses or visit reasons, the associations between diagnoses and benzodiazepine use were virtually unchanged. Third, NAMCS does not account for whether a prescribed medication is taken regularly versus as needed, so it is possible that the extent of use is overestimated. However, because only eight medications are recorded, benzodiazepine use may also be underestimated. Fourth, because NAMCS is a survey of office-based practice, it does not include physicians practicing in other settings. Although physician nonresponse might introduce bias into the results, the survey weights that NAMCS designed account for this to produce unbiased national estimates.[24] Finally, although the analysis was limited to nonpsychiatrist physicians, it is possible that psychiatrists initially prescribed some of the continuation benzodiazepines. Nevertheless, because nearly 95% of new benzodiazepines prescribed to older adults are from nonpsychiatrists,[1] it is likely that the psychiatrist-initiated group accounts for a small subset.

These nationally representative analyses largely confirm and update analyses using data from more than 20 years ago by demonstrating that benzodiazepine initiation continues into late life, continuation use increases with age, and benzodiazepines are prescribed for purposes other than clearly defined mental disorders.[22,40] How can continued use of a potentially harmful intervention, in the face of extensive evidence that alternatives are effective and safer, be explained? A previous study[13,14] of older adults who were chronic benzodiazepine users suggested that people doubted that "[any]thing other than the benzodiazepine" would help and generally rejected the idea of psychological interventions. The majority of physicians "believed that attempting withdrawal would be time-consuming and likely futile." These attitudes are frustrating given the growing evidence that older adults, even those with chronic use, can successfully decrease and be tapered off benzodiazepines using interventions including cognitive behavioral therapy and direct-to-consumer educational techniques.[17–19,41] As attitudes about mental health disorders and treatment change, it may be that older adults will become more willing to consider psychotherapeutic treatment options. However, this will not be helpful if they have no access to specialty mental health services.[42]

Although people may be reluctant to stop long-term benzodiazepine use, a physician's fundamental responsibility is to the safety of his or her patient. The majority of use appears to be in the absence of a clearly defined mental disorder, with limited use of alternative and safer treatments such as antidepressants or psychotherapy. Although clicking "reorder" may limit short-term patient (and provider) distress, there are critical concerns regarding the appropriateness and safety of most long-term benzodiazepine use in older adults in the United States. New strategies are needed to encourage patients and providers to discontinue potentially inappropriate benzodiazepine therapy.