Do benzodiazepines contribute to respiratory problems?

Abstract

Non-selective benzodiazepines are a class of sedative and anxiolytic medication that are commonly prescribed. Physiology studies and animal studies suggest that non-selective benzodiazepines may adversely impact respiration through a variety of mechanisms. Several recent, well-designed, population-based observational studies confirm that benzodiazepine-related negative respiratory outcomes are a concern. In this article, the mechanisms and clinical evidence for non-selective benzodiazepine-related adverse respiratory outcomes, as well as the methodological issues relating to the evaluation of adverse drug effects are reviewed.

Keywords::

• benzodiazepines
• chronic obstructive pulmonary disease
• epidemiology
• pneumonia
• sedatives

Benzodiazepine drugs are all too frequently used. In the general US population, prevalent non-selective benzodiazepine drug use in 1987 was estimated to be about 6% (10.9 million people) [1]. Among older adults, who are at increased risk for benzodiazepine drug side-effects, prevalent use in North America in the 1990s was estimated to be even higher, at 10–20% [2,3]. Among individuals with chronic obstructive pulmonary disease (COPD), which is another subgroup likely at increased risk for benzodiazepine-related adverse effects, the use of this drug is even more common, at an estimated 30% [4]. Non-selective benzodiazepine use may also be increasing over time [5], although there are conflicting data regarding this [1,3,6]. Besides being commonly prescribed, non-selective benzodiazepines are also used in other potentially concerning ways, such as prolonged duration of use [4,7], repeated periods of use within a single individual [4,7], high dosage [7] and concomitant use of other psychoactive [2,7] and narcotic [8] medications. Among individuals with COPD, particularly among those with more severe disease, non-selective benzodiazepines are also used during periods of respiratory exacerbation [4].

Physiology studies and animal studies suggest that non-selective benzodiazepine drugs may cause respiratory harm through multiple possible mechanisms. First, they may cause harm via the central nervous system, which is abundant with GABA_A receptors, to which benzodiazepines bind. Benzodiazepine administration has been shown to depress central respiratory drive and chemoreceptor responsiveness to hypercapnea [9,10]. Benzodiazepines also decrease arousability from sleep [11], which may prolong the duration of hypoventilation or contribute to aspiration. Second, non-selective benzodiazepines may directly influence the pulmonary system by activating GABA_A receptors that are located in the peripheral nervous system or in peripheral tissue [12]. For example, benzodiazepine drugs have been shown to decrease inspiratory and expiratory respiratory muscle strength [13]. Third, non-selective benzodiazepines have been reported to exert effects on the lower esophageal sphincter. Benzodiazepines decrease lower esophageal sphincter pressure [14], perhaps through the activation of peripherally situated GABA_A receptors [12], which can then facilitate reflux, and possibly aspiration. Finally, because GABA_A receptors are also found on monocytes and macrophages, non-selective benzodiazepines may exert immune system effects. Benzodiazepines have been reported to provoke intracellular acidosis in alveolar macrophages, leading to impaired cytokine production and bacterial phagocytosis and killing [15].

While physiology studies and animal studies offer insights into mechanisms by which benzodiazepines may cause respiratory harm, they offer limited information on the actual clinical impact of these drugs in human populations. Although randomized controlled trials of drug efficacy often monitor for medication-related adverse events, these studies are limited in their ability to comprehensively evaluate harm, as they usually involve relatively small numbers of participants, selected individuals, short durations of follow-up and limited drug dosing. Randomized controlled trials of drug harm are generally difficult to undertake because of ethical concerns. Population-based observational studies are well-suited to evaluate drug-related adverse outcomes for several reasons. First, observational studies invariably include larger numbers of participants than randomized controlled trials, which facilitate identifying potential drug-related adverse outcomes that are likely to occur in a minority of individuals. Second, individuals from the broader population are usually included in observational studies, like older adults and individuals with comorbidities, in whom drug-related adverse events are more likely to occur. Such groups of individuals are often purposefully excluded from randomized control trials. Third, relatively longer durations of follow-up are generally possible in observational studies, facilitating more comprehensive identification of adverse events that may not necessarily occur early on following treatment initiation. Finally, observational studies are able to evaluate ‘real-world’ dosing and use of drugs.

Three recent, well-designed, population-based observational studies provide evidence that non-selective benzodiazepines are associated with clinically important adverse respiratory outcomes. Using UK population-based health administrative data, Obiora et al. [16] evaluated benzodiazepine drug exposure in relation to doctor-diagnosed pneumonia using a case–control design, as well as all-cause mortality following pneumonia using a retrospective cohort design. This study included individuals from the general population (without age or comorbidity restrictions), and analyses were adjusted for a broad range of variables, including, but not limited to, pneumonia pre-study period, specific cardio-pulmonary comorbidities and smoking. Benzodiazepine drug exposure was associated with a 54% significantly increased adjusted risk of pneumonia and a 32% significantly increased adjusted risk of all-cause mortality. Increased risks of pneumonia and mortality were also found in association with receiving various specific individual benzodiazepine drugs, including, of note, the selective benzodiazepine receptor drug, zopiclone. Findings persisted among the subgroup of individuals with no comorbidities, which is the group least likely to be influenced by confounding by indication. Using population-based health administrative data from Ontario, Canada, and a retrospective cohort design, Vozoris et al. [17] examined incident non-selective benzodiazepine drug exposure and various adverse respiratory outcomes in a respiratory-vulnerable group, older adults with validated doctor-diagnosed COPD. After propensity score matching on 26 different covariates, benzodiazepine users were found to have a 45% significantly increased risk of outpatient respiratory exacerbations and 92% significantly increased risk of emergency room visits for COPD or pneumonia. Similar to the results of Obiora et al. [16], increased risks of adverse respiratory outcomes were found among users of short/intermediate- and long-acting benzodiazepines, and also among individuals with less severe COPD, which again is the subgroup least likely to be influenced by confounding by indication. Finally, using Swedish population-based health administrative data and a prospective cohort design, Ekström et al. evaluated benzodiazepine use among individuals with doctor-diagnosed COPD newly receiving supplemental home oxygen [8]. Although outcomes were not respiratory-specific (i.e., all-cause hospitalization and all-cause mortality), it is reasonable to assume that many of the hospitalizations and deaths were probably respiratory-related given that a COPD population with an advanced degree of disease was being studied. Analyses were adjusted for a broad range of relevant covariates including, but not limited to, WHO performance status, forced expiratory volume in 1 s, arterial blood gas oxygen and carbon dioxide tensions, smoking status, number of previous hospitalizations and comorbidities. While benzodiazepines were not associated with all-cause hospitalization, they were associated with a 21% significantly increased adjusted risk of all-cause mortality, and a dose-dependent relationship was also demonstrated.

Although the finding of clinically important adverse respiratory outcomes in association with non-selective benzodiazepine use in these large population-based studies is impressive, there are limitations that should also be acknowledged. First, causation cannot be inferred in observational study designs. Second, although a fairly extensive list of confounders was included in these aforementioned studies, there is always the possibility that unresolved residual confounding accounts for the observed associations between benzodiazepine exposure and adverse respiratory events. One specific confounder that is not completely controlled for in any of the above studies is that of drug indication. It is possible that benzodiazepine recipients received their medication to help treat pre-existing respiratory symptoms (like dyspnea), or anxiety related to pre-existing respiratory symptoms, and that the subsequent diagnosis of pneumonia or respiratory exacerbation is unfairly linked with the benzodiazepine. While the aforementioned studies attempted, in part, to address confounding by indication by, for example, adjusting analyses by history of previous pneumonia or respiratory exacerbation or by examining outcomes in the healthiest subgroup of individuals, these approaches do not completely overcome the specific limitation. Third, because the above studies were based, in part, on analysis of health administrative data, the clinical validity of some variables and measures may be called into question. Finally, it should be noted that there are other observational studies demonstrating no positive association between non-selective benzodiazepines and adverse respiratory health. Dublin et al. found no significantly increased risk of pneumonia (identified by validated health administrative data) among older adult benzodiazepine recipients, but this study included a much smaller number of participants compared to those previously mentioned, and all the participants belonged to one health insurance company, potentially limiting generalizability [18]. Almirall et al. found that benzodiazepines were actually protective against clinically diagnosed pneumonia in a large general population sample, but drug exposure was based on patient report, rather than a more objective assessment [19]. Iqball et al. reported no increased risk of pneumonia with benzodiazepine receipt using Taiwanese population-based health administrative data, but many methodological details relating to this analysis are not described, including which covariates were incorporated and how [20].

The finding of substantially increased risk of clinically important adverse respiratory outcomes in association with non-selective benzodiazepine receipt in multiple, large, population-based analyses from different parts of the world and the persistence of results across individuals with differing baseline health support the conclusion of a true link between benzodiazepine use and respiratory harm. Results of physiology studies and animal studies provide a supportive biologic platform for the observed associations. Therefore, the potential for adverse respiratory effects needs to be taken into consideration when prescribing non-selective benzodiazepines, especially for vulnerable subgroups like older adults and individuals with COPD [8,17]. To know whether the risk for adverse respiratory outcomes extends to other specific psychoactive drug categories, such as selective benzodiazepine receptor drugs, tricyclic antidepressants, serotonin reuptake inhibitors etc., requires further research. The fact that there is research emerging linking opioid agonists with adverse respiratory outcomes [8,18] suggests that the problem may be wider than just non-selective benzodiazepine drugs, potentially involving sedating medications more broadly.

Financial & competing interests disclosure

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.