Association of benzodiazepine and Z-drug use with the risk of hospitalisation for fall-related injuries among older people: a nationwide nested case–control study in Taiwan

Falls are common in older people and are the second leading cause of accidental or unintentional deaths worldwide [1]. Approximately 28%–35% of people aged 65 years and older fall each year [2], and a study found that fall-related trauma accounts for 5.3% of all hospitalisations in this age range [3]. The risk of falls is multifactorial, and medications are modifiable extrinsic risk factors [4, 5]. Meta-analyses and systematic reviews have reported that some classes of medications, such as benzodiazepines (BZDs), increase the risk of falls in older people [6‐8].

BZDs are among the most prescribed psychotropic medications, especially in older people [9]. Because BZDs can be used as sedatives and hypnotics, they can cause problems as dependence and abuse [10] besides the side effects of dizziness, drowsiness, and coordination impairment. Moreover, after controlling the confounding factors, several studies have reported that BZDs independently increase the risk of accidents such as falls [11, 12], hip fractures [13, 14], and motor vehicle accidents [15]. However, data on whether different characteristics (as exposure duration, daily dose, and elimination half-life) lead to different risks are inconsistent. A study reported that the use of long-acting BZDs increases the risk of falls [16], whereas other studies have reported that the use of short-acting BZDs also increases this risk [17, 18]. In addition, another study indicated that dosage contributes more to the risk of falls than elimination half-life [19].

Z-drugs (zolpidem, zopiclone, and zaleplon) are non-BZD hypnotics and are advocated to be safer than BZDs [20]. Thus, the prescription of Z-drugs has been increasing rapidly [21]. A study reported that the annual use of Z-drugs in older people doubled from 2001 to 2010 in Taiwan [22]. However, some studies have demonstrated that Z-drugs also cause problems of abuse and dependence [23], and increase the risk of falls [24] and hip fractures [25].

Polypharmacy is an arising issue in public health and has a crucial role as a risk factor for falls [26]. Studies have found that the simultaneous use of two or more anxiolytics or hypnotics (another kind of polypharmacy) is common in Taiwan [27]. However, whether such polypharmacy with BZDs and Z-drugs increases the risk of fall-related injuries requiring hospitalisation in Taiwan remains unknown.

Since it was found that BZDs and zolpidem (one kind of Z-drugs) were two of the top five reported abused drugs by medical institutions in Taiwan [28] and that their long-term use (defined as 180 prescription days within a year) was not uncommon [29], it is important to understand their potential risk. This study investigated the association of the use of BZDs and Z-drugs with the risk of hospitalisation for fall-related injuries, with a focus on exposure duration, daily dose, elimination half-life, and BZD and Z-drug polypharmacy in older people.

The demographic and clinical characteristics of the case and comparison subjects are listed in Table 1. The mean age of the fall patients was 77.7 years, and 60.7% were women. Compared with the control subjects, the case subjects were more likely to have been diagnosed with dementia, Parkinson’s disease, and cerebrovascular disease while controls were more likely to have been diagnosed of hypertension and ischemia heart disease during the 365 days before the index date. In addition, the case subjects were more frequently exposed to antipsychotics, antidepressants, and opioids but less likely to exposed to thiazide diuretics during this study period. Compared with the control subjects, the case subjects had significantly less received outpatient services in the 180 days before the index date.

Table 2 illustrates the relationship between the risk of hospitalisation for fall-related injuries and the types of exposure to BZDs and Z-drugs. The current users of BZDs were significantly associated with a high risk of hospitalisation for fall-related injuries after adjustment for covariates (AOR = 1.32, 95% CI = 1.17–1.50). However, the risk was lower in the indeterminate and former users of BZDs than in the nonusers (AOR = 0.75, 95% CI = 0.62–0.91 for indeterminate users and AOR = 0.74, 95% CI = 0.64–0.86 for former users). Compared with Z-drug non-users, the current users of Z-drugs also showed significantly higher risk of hospitalisation for fall-related injuries after adjusting the covariates (AOR =1.24, 95%CI = 1.05–1.48).

Table 3 demonstrates the relationship between the risk of hospitalisation for fall-related injuries and the dose levels of BZDs and Z-drugs among the current users. After adjustment for covariates, significantly increased AORs were observed for all dose levels of BZDs users (AOR = 1.75, 95% CI = 1.47–2.08; AOR = 1.54, 95% CI = 1.28–1.85; and AOR = 1.27, 95% CI = 1.08–1.50, respectively, for high, medium, and low dose levels). For the current users of Z-drugs, only the high dose level significantly increased the risk after adjusting the covariates (AOR = 1.37, 95% CI = 1.14–1.64).

Table 4 shows the relationship between the risk of hospitalisation for fall-related injuries and the characteristics of BZD exposure with respect to the index date, the elimination half-life, and polypharmacy. For the elimination half-life, all three categories, namely long-acting BZDs (AOR = 1.41, 95% CI = 1.16–1.71), short-acting BZDs (AOR = 1.42, 95% CI = 1.20–1.69), and combined long- and short-acting BZDs (AOR = 1.61, 95% CI = 1.37–1.89), significantly increased the risk after adjustment for covariates. Compared with nonusers, treatment with only one kind of BZD and only Z-drugs were associated increased risk of fall-related injury (AOR = 1.40, 95% CI = 1.19–1.65 and AOR = 1.33, 95%CI = 1.04–1.69). Regarding polypharmacy, two or more types of BZD (AOR = 1.61, 95% CI = 1.38–1.89), one kind of BZD+ Z-drugs (AOR = 1.65, 95% CI = 1.08–2.50), and two or more types of BZD plus Z-drugs (AOR = 1.33, 95% CI = 1.04–1.69) significantly increased the risk of hospitalisation for fall-related injuries. The risk of fall-related injury requiring hospitalisations when polypharmacy were all significantly more prominent than monotherapy of only one kind of BZD or only Z-drugs alone.

In this population-based study, we observed that the current use of BZDs was associated with an increased risk of fall-related injuries requiring hospitalisation in older people, irrespective of dose levels (low, medium, or high) and elimination half-life (long-acting or short-acting). By contrast, the current use of Z-drugs, in particular at high dose levels increased this risk. With respect to polypharmacy, the use of more than one type of BZD significantly increased the risk, which reached the highest prominence when two or more types of BZD plus Z-drugs were used.

Our study results revealed that even low doses (<0.3 DDD) of BZDs were associated with an increased risk of fall-related injuries requiring hospitalisation among older people, and this risk correlated with an increase in the dose. A study [19] reported that a BZD daily dose higher than 0.75 DDD increased the risk of falls leading to femur fractures among subjects aged 55 years and older. The results of our previous study [13] and another study [14] demonstrated that a BZD daily dose higher than 0.3 DDD increased the risk of hip fractures in older people. These studies suggest that physicians should avoid prescribing BZDs to older people considering the high risk of falls even on low dose.

The 2002 Beers criteria [34] suggested that physicians should avoid prescribing long-acting BZDs to adults aged 65 years and older because they are potentially inappropriate medications. However, the 2012 and 2015 Beers criteria [35, 36] further suggested that the prescription of any BZD to older people is potentially inappropriate. Our data are consistent with those of previous studies [13, 14, 18] and support the notion that short-acting BZDs are not safer than long-acting BZDs in older people.

Previous studies have reported that zolpidem increased the risk of falls [24] or fractures [25] in older people, our results also indicated that the current use of Z-drugs, in particular at high dose levels (>0.6 DDD) increased the risk of hospitalisation for fall-related injuries in older people.

In this study, we observed that the concomitant use of two or more types of BZD or any BZDs plus Z-drugs were significantly associated with the risk of hospitalisation for fall-related injuries among older people. Such definition of polypharmacy may differ from other studies [26, 37], which define it as the combined use of 5 or more drugs daily. Because such anxiolytic-hypnotic polypharmacy is common and has increased in Taiwan [27], more attention should be focused on the potential risk of falls in older people.

Two unexpected findings were noted in this study. First, the cases were less numbers of outpatient services in the 180 days before the index date. It is possible due to a bias in the selection of controls (e.g., some controls may have severe physical comorbidities which make them totally bed ridden and less prone to falls). Second, compared with BZDs non-users, indeterminate users and former users seemed to have lower risk of fall-related injuries than non-users. Such results may be due to some uncontrolled confounding factor. However, the indeterminate and former users may have protected effects which suggest older people who stop BZDs may reduce the risk of fall-related hospitalisations. Several limitations should be considered before interpreting our results. First, since the patients with less severe fall-related injuries may not have been recorded in the NHIRD, we used E-code to define the outcome. However, the E-code may not routinely be recorded by physicians in Taiwan and may have poor sensitivity. Such definition may underestimate or overestimate the fall risk of BZDs / Z-drugs to the elderly. Second, we defined BZDs and Z-drugs exposure within 1–30 days before the index date as “current use”, which followed the classification of Ray et al. [31] in their earlier paper. However, they proposed the potentially serious misclassification in the study of the “acute effects” of BZDs and other drugs used intermittently upon studies of related injuries and suggested to track exposure on a day-by-day basis in future studies [38]. Third, although we used CCI and healthcare utilizations to control the comorbidities, some additional residual confounding factors should be considered such as vision problems, body mass index, physical activity, smoking, and alcohol use which may not be available from our data. Additionally, the functional abilities (gait speed, difficulties with activities of daily living) would be also important but not be available with administrative data. Forth, because the NHIRD only provided prescription records of drugs, we are unable to assert the exactness of each subject’s medicating status, as noncompliance or “on a need basis” are also possible. Last, confounding by indication is another major bias in the observatory pharmacoepidemiological studies and may underestimate or overestimate the risk between BZDs/Z-drugs and fall in elderly people.

Despite several limitations listed above, our study has several strengths. The sample size used (NHIRD) was large, as it consists of a nationwide registry for medical claims data in Taiwan. In addition, ours is one of the few studies simultaneously examining the specific characteristics of BZD and Z-drug use among older people in a single study, with a focus on the type of exposure, daily dose, elimination half-life, and polypharmacy. Additionally, this is the first study to investigate the effects of polypharmacy with BZDs and Z-drugs on the risk of hospitalisation for fall-related injuries among older people. Consideration that Z-drugs are increasingly used as a common treatment choice for insomnia and the combined use of BZDs and Z-drugs is increasing in older people, the focus of this study is relevant for clinical decisions.

In conclusion, this study shows that different dose levels and half-lives of BZDs, a high dose of Z-drugs, and polypharmacy with BZDs and Z-drugs were associated with an increased risk of fall-related injury requiring hospitalisation in older people. Physicians should balance the risks and benefits when prescribing these drug regimens to older people considering the risk of falls.