High-Dose Benzodiazepine Use among Long-Term Users: When Will We Ever Learn?

Abstract

In this issue, Cloos et al. [1] present a population-based study on high-potency benzodiazepine (BDZ) use in Luxembourg, a nation with universal coverage, thus allowing a longitudinal analysis of utilization across the years 1995 through 2007. The study categorized those with one or more BDZ dispensings according to the duration of exposure, namely: (1) short-term (less than a 3-month quantity); (2) intermediate (multiple dispensings with at least a 1-year interruption), and (3) continuous use (>3-month quantity with no interruption). In addition, a second characteristic, high-dose use (HDU), was examined within each exposure group. HDU was defined as exceeding the usual therapeutic maximum daily adult dose as recommended by major compendia, i.e. Martindale and Micromedex.When BDZ anxiolytic and hypnotic users were identified according to the use of BDZ for anxiety or insomnia and combined across the two drug groups, the data revealed that:• Among total Luxembourg enrollees, 16% had one or more BDZ dispensings yearly, with females being 38% more likely than males to have a BDZ dispensing (unadjusted prevalence ratio). • Overall, among the three exposure groups, 18.4% were continuous users, although the female-to-male difference was less pronounced in this group.• HDU grew according to the length of exposure such that HDU was less than 1% among low-exposure use, 3.4% in intermediate exposure while HDU was 20.3% among continuous users. The data support the authors' decision to conduct subsequent analysis targeting the group most at risk, i.e. the 39,324 continuous users wherein HDU was most prevalent.In the subsequent analysis of continuous users, a multivariable logistic regression using a generalized estimating equations analysis was undertaken to account for persons with repeated BDZ dispensings across the years, stratified according to age at first BDZ intake. The authors present the adjusted odds separately for drugs primarily used as anxiolytics and those primarily used as hypnotics. It should be noted that in the absence of diagnostic information, there are proportions of outpatient use for seizure disorders and other nonpsychiatric conditions which presumably did not vary across the years. Finally, the reference group was diazepam for both anxiolytics and hypnotics. The rationale for the choice of diazepam as the comparator presumably relates to its medium potency, rapid onset and long length of action to avoid the problems high-potency, short-duration products have presented in terms of difficulty in withdrawal, dose escalation and dependence [2]. The analysis revealed that:• Among anxiolytics, only two products had significantly increased odds of HDU compared with reference diazepam, and both of these occurred among the 69- to 105-year-olds at first BDZ intake: alprazolam and prazepam. The increased risk with prazepam is not consistent with the theory that high-potency, short-acting products are most likely to produce HDU, but could be an artifact. • By contrast, hypnotics showed a strikingly different risk assessment: The odds of HDU compared with diazepam were significantly greater regardless of the age at first BDZ intake. In the 69- to 105-year-olds, 7 of the 8 products assessed had significantly greater odds of HDU than diazepam, the reference compound. Greatly increased odds were found for both brotizolam and triazolam with HDU twice as common among triazolam users. Study data on the triazolobenzodiazepines corroborate prior findings that short-acting high-potency products are particularly prone to long-term use problems, e.g. withdrawal syndrome and dependence. These can lead to serious adverse events, e.g. severe anterograde amnesia [3]. The data add support for recommendations for clinical guidelines and health payment systems to restrict inappropriate use. In general, the strengths and limitations of administrative claims data are carefully documented by the authors. The opportunity to examine data longitudinally over 12 years in a large cohort of users is particularly impressive. More information on the distribution of BDZ days of use among continuously exposed persons would have added further strength, and addressing the extent of HDU as a function of both age and continuity of use adds a novel method to assess these clinically important patterns of use.Of course, the external validity of data in a national health registry gives great credence to the implications for clinical education of practitioners and patients and for enhanced oversight by payers. For example, the sudden growth of antipsychotic use in very young children for nonpsychotic conditions, e.g. attention deficit hyperactivity disorder, has prompted numerous state Medicaid systems to require peer review of requests for such use [4]. In terms of improving clinical prescribing behavior, while HDU appears relatively restricted in these Luxembourg population data, avoidance of both long-term anxiolytic and hypnotic use in elder