Why Z-drugs are used even if doctors and nurses feel unable to judge their benefits and risks—a hospital survey

Many patients receive Z-drugs for hospital-associated sleep problems, in spite of well-known risks. The aim of this study was to learn more about the attractiveness of Z-drugs, seen from the doctors’ and nurses’ perspective. Using a standardized questionnaire, doctors (63/116) and nurses (73/243) in a German general hospital were surveyed about the risks and benefits of Z-drugs, compared with benzodiazepines. “Reduced time to get to sleep” was perceived by doctors (51%) and nurses (53%) to be a strong benefit of Z-drugs; “confusion” and “falls” were perceived by ca. 10% of doctors and ca. 15% of nurses to be a frequent problem. Compared with benzodiazepines, respondents more often answered “unable to judge” for Z-drugs; e.g. for doctors, 18% (benzodiazepines) vs. 45% (Z-drugs) were unable to judge “improved daytime functioning” and 12% (benzodiazepines) vs. 37% (Z-drugs) were unable to judge “falls.” Z-drugs seem to be attractive because experiential knowledge overemphasizes their benefits and fails to take risks such as drug-related falls and confusion into account. Difficulties to judge a drug’s risk-benefit ratio do not prevent doctors and nurses from using them. Interventions for reducing Z-drug usage should incorporate local quality assurance data about relevant patient risks.


Introduction
Many hospital doctors experience a conflict every night when on duty: what to do with patients who have trouble sleeping? For severe cases of chronic insomnia, cognitive behavioral therapy and hypnotic drug treatments are recommended. However, transient sleep problems in the hospital-often linked to environmental factors such as unfamiliar sounds, nursing interruptions, uncomfortable beds, and bright lights [1]-are different from a clinical diagnosis of insomnia disorder, which affects sleep onset, duration, and/or quality for at least a month [2]. Guidelines on how to treat transient sleep problems are not helpful in the hospital environment so that doctors and nurses are challenged to manage inpatient sleeping problems.
Chart review studies give us an idea of how hospital doctors usually solve this conflict; they often prescribe benzodiazepines and newer non-benzodiazepines (so-called Z-drugs) for patients who have trouble sleeping [3][4][5]. While these drugs may help patients to sleep in the hospital environment, they also have adverse effects, such as confusion, falls, fractures, and craving [6] so they are not recommended for the treatment of transient sleep problems in most guidelines [7]. Z-drugs, specifically, have been accompanied with conflicting information ranging from "[Z-drugs are] considered the safest and most effective prescription sleep aids for geriatric patients" [8] to "[Z-drugs are] not necessarily a safer alternative to traditional BZDs" [9].
The goal of this study was to understand why Z-drugs remain an attractive solution for doctors and nurses to transient sleep problems. We performed a survey to learn more about how hospital doctors and nurses perceived Z-drugs compared with benzodiazepines. After an initial review of the data of this survey, we became aware that many doctors and nurses in our sample checked "unable to judge" when answering items about the benefits and risks of Z-drugs. Therefore, we were also interested in factors associated with the inability to judge the risks and benefits of Z-drugs.

Method
Context: the sleeping pills project In a mixed-methods project, we strived to add knowledge about the current use of sedatives and hypnotics in hospital and primary care from multiple perspectives and with several different types of data [10]. For example, a hospital chart review showed that 12% of all patients 65 and older received a Z-drug at least once during their stay [11]. A patient survey in the same hospital revealed that more than half of the older patients who received Z-drugs in the hospital wished to continue taking these drugs at home [12].
The data reported here come from a survey from the professional perspective performed between June and September 2014. The study was approved by the Ethics Committee of Göttingen University Medical Center (25/2/14). Parts of the survey results have already been analyzed and published [13,14], with a focus on benzodiazepines.

Design, participants, and measures
In a cross-sectional survey, all physicians and nurses of a German general hospital received a paper questionnaire about the risks and benefits of benzodiazepines and Z-drugs. The questionnaire was developed by Siriwardena et al. [15] to explore general practitioners' beliefs about the benefits and risks of hypnotic prescribing. Respondents are asked to rate the extent of different benefits (on a 5-point scale, ranging from "very small" to "very strong") and the frequency of risks (from "never" to "always") for benzodiazepines and Z-drugs. Both scales have an additional answer category, "unable to judge." Hoffmann translated this questionnaire into German to survey German general practitioners [16]. We used Hoffmann's translation in this study, making slight adjustments to the questionnaire for use in a hospital setting.

Statistical analysis
The relative frequencies of doctors' and nurses' assessments of the six benefits and five risks of Z-drugs and benzodiazepines were analyzed. We combined the two extremes "strong/ very strong" for the benefits and "often/always" for the risks of both Z-drugs and benzodiazepines. We compared doctors' and nurses' perspectives about "strong benefits" and "frequent risks" for both Z-drugs and benzodiazepines.
We then explored in more detail the group of respondents who checked the "unable to judge" box when asked about the risks and benefits of Z-drugs. First, we compared the percentage of doctors and nurses who answered "unable to judge" to the items about Z-drugs with the percentage who answered "unable to judge" for benzodiazepines. In a second step, we divided the sample into those who were able to judge the majority of items about Z-drugs (i.e., 0-5 unable to judge answers) versus those who were not (i.e. 6-11 unable to judge answers). In a multivariable logistical regression analysis, we modeled the likelihood for being unable to judge Z-drugs, taking the following five factors into account: sex, profession, length of employment, department, and self-reported frequency of use. We report adjusted odds ratios (ORs) and their 95% confidence intervals (CI). All analyses were performed using SAS 9.4.

Results
More than half of the doctors (63/116) and about one-third of nurses (73/243) participated in the survey. Most doctors were male (66%), about one-third worked either at departments of internal medicine or surgery, half of them for less than 5 years. Nearly 81% of the responding nurses were female; the majority had more than 10 years working experience (Table 1).
With regard to the benefits of Z-drugs, most respondents appreciated "reduced time to get to sleep" (with 51% of the doctors and 53% of the nurses reporting a strong/very strong benefit; Table 2) and "reduced night-time waking" (41% doctors, 63% nurses). Nearly as many doctors reported these advantages for benzodiazepines, but fewer nurses (Table 2). Both groups clearly favored benzodiazepines for "reducing fear or agitation." We also found common features of, and differences between, the perceived risks of Z-drugs and benzodiazepines. The order of risks for Z-drugs was nearly the same for benzodiazepines, but far more doctors perceived them as occurring less frequently in Z-drugs than benzodiazepines (Table 2). This difference did not apply to nurses. Both doctors and nurses agreed about the most frequent risk for Z-drugs, 38% of doctors and 45% of nurses rated "craving" as occurring frequently (Table 2). Only few respondents found "confusion" (10% doctors, 15% nurses) and "falls" (10% doctors, 16% nurses) to be frequent problems of Z-drugs.
Interestingly, far more doctors and nurses answered "unable to judge" for the benefits and risks of Z-drugs in comparison with benzodiazepines. For example, 45% (Z-drugs) versus 19% (benzodiazepines) of doctors answered unable to judge for "improved daytime functioning" and 37% (Z-drugs) versus 12% (benzodiazepines) answered unable to judge for "falls." Similarly, more nurses answered unable to judge "falls" for Z-drugs (18%) than for benzodiazepines (8%).
A total of 33 of the 138 respondents (24%) answered "unable to judge" for 6 or more of the 11 items about Z-drugs compared with 14 respondents (10%) for benzodiazepines. This difference is significant (Fisher exact test, OR = 2.77, p = .002). A multivariable logistic regression showed that sex did not play a role in the ability to judge the benefits and risks of Z-drugs (Table 3). However, four other characteristics (profession, length of employment, department, and selfreported frequency of use) had a significant effect upon the likelihood of being able to judge Z-drugs. The strongest predicting factor was being a doctor (adjusted OR 7.79; 95% CI 1.85 to 32.56), followed by infrequent use of Z-drugs on one's ward (3.99; 1.17 to 13.60).

Discussion
Many respondents in our survey appreciated the potential of Z-drugs to reduce the time to get to sleep; especially nurses saw a strong effect in reducing night-time waking. For many doctors and nurses, Z-drugs are thought to cause less confusion and fewer falls than benzodiazepines. This is surprising since a Canadian meta-analysis of risks and benefits of short-term treatment with sedative hypnotics in older people with insomnia concluded already 14 years ago that the benefits of both benzodiazepines and Z-drugs are only marginal and outweighed by the risk of falls or cognitive impairment, particularly in a high-risk elderly population [17]. Later studies confirmed these results, most recently a review from Canada in which Lee and colleagues [18] plea for a strategy of deprescribing benzodiazepines and Z-drugs for insomnia. How, then, did the doctors and nurses in the hospital under study come to their positive risk-benefit assessment of Z-drugs?  First of all, they do not stand alone with their positive assessment of Z-drugs. Many surveyed general practitioners also perceived Z-drugs as more effective and safer compared w i t h b e n z o d i a z e p i n e s [ 1 6 ] . I n a d d i t i o n , pharmacoepidemiological studies report a reduction of benzodiazepine prescribing while Z-drug prescribing is increasing [19]. Z-drugs still seem to benefit from the myth that they are similarly effective and safer alternatives to benzodiazepines, although studies show that Z-drugs also carry the same risks of daytime sedation, cognitive impairment, falls, fractures, and accidents [18].
Second, since Z-drugs are hypnotics, i.e., prescribed primarily to treat sleeping problems, they may not have the same value in the eyes of hospital doctors as benzodiazepines, which are prescribed for several other indications in addition to sleeping problems. Interview studies with hospital doctors have shown that doctors regard other medical issues with higher priority than sleep problems during hospitalization [20].
Third, based on meta-analytic data on sleep efficacy reported mostly by younger patients, Rösner et al. [21] could show a 12-min decrease of sleep onset latency, a 17-min decrease of wake time after sleep onset, and a 28-min increase of total sleep time for eszopiclone, compared with placebo. Although the possible benefits of Z-drugs are relatively minor, it seems that it is exactly these minor effects that matter: for doctors and nurses working in night shifts under high stress and for patients desperately seeking a good night's sleep in a busy hospital and unfamiliar environment [20].
For a deeper understanding why individuals prefer some risky choices over other options, such as Z-drugs to combat hospital-associated sleep problems, a look at recent research on the role of descriptive knowledge (e.g., official drug information) and experiential knowledge (e.g., seeing benefits and risks of a drug first or second hand) in decision making may be helpful. Research investigating how descriptive and experiential knowledge affect choices between risky options found that people give more weight to experiential knowledge [22,23]. That is, they are more influenced by the perceived outcomes of their own choices rather than by accurate summary statistics about the outcomes across many decisions taken by many people. Importantly, people rely on experiential knowledge even when it is limited and based on rather small samples [22,23]. When deciding on Z-drugs, doctors and nurses may also have observed only small samples. For example, data analyses show that "falls" occur more often in older hospital patients treated with Z-drugs [24], but falls themselves are rare adverse events and, therefore, difficult for a doctor or nurse to observe. Even if doctors and nurses were taught differently via descriptive knowledge, the experiential knowledge of regularly giving Z-drugs and seeing no direct harm may lead to the belief that Z-drugs do not increase the risk of falls. Vice versa, doctors who were convinced, for whatever reason, that "falls" and/or "craving" are a frequent adverse effect of benzodiazepines, believed the risks outweigh the benefits of these drugs [14]. A second result of our survey was no less surprising: despite the fact that 12% of older patients are treated with a Zdrug in this hospital [11], a considerable group of respondents checked the unable to judge answer category for the majority of questions about the risks and benefits of Z-drugs. The above-mentioned description-experience approach for explaining risky choices [23] may also help to explain this result. We know from a chart review in this hospital [11] that Z-drugs are much less commonly used in surgical departments. Consequently, this lack of experience with Z-drugs on surgical departments is a logical explanation why fewer surgeons and nurses on surgical wards were able to answer survey questions about the benefits and risks of Z-drugs.
Most interesting is the strong difference between doctors and nurses in the ability to judge Z-drugs, in spite of the fact that the majority of respondents should have had some contact with Z-drugs (experiential knowledge) during the course of their professional careers. All six benefits and, at least, three of the unwanted drug effects ("confusion," "falls," and "craving") can be directly observed due to the fact that these effects come about (or not) within hours of administering the drug. Although doctors are responsible for diagnosing diseases and prescribing medicines, they often prescribe Z-drugs as p.r.n. (pro re nata; as needed) drugs. Since the "as needed" case often occurs during the night shift, doctors may lack direct experience with the patients who use Z-drugs. Nurses, on the other hand, have direct patient contact day and night and are able to directly observe the benefits and adverse effects of these drugs. This difference in experience may explain why nurses are better able to judge the benefits and risks of Z-drugs, especially for items like "reduced time to get to sleep," which are immediate, directly observable results of Z-drug use.

Strengths and limitations of the study
The most important limitation of the study is the relatively small sample size. The results of a single-hospital study can only be a starting point for future research and are not-of course-generalizable for other settings.
A strength of the study is that we were able to compare within the setting of a single hospital how doctors and nurses perceived and assessed Z-drugs and the frequency with which these drugs are prescribed [11]. However, quantitative analyses of survey data cannot explain the underlying reasons for behavior. Rather, qualitative data (e.g., interviews) may provide an explanation for the discrepancy between unable to judge survey responses and actual Z-drug prescription practices.
An additional strength of the study is the inclusion of both the doctor and nurse perspectives, since the management of hospital-associated sleep problems is a multi-professional task and any change of the current practice will need multiprofessional efforts. A high percentage of the hospital's doctors participated in the study, even though doctors' willingness to participate in surveys is generally limited. Given the leading role of hospital doctors in a hospital's prescribing policy, the high response rate contributes to the internal validity of the study results. We can only speculate why two-thirds of nurses declined to participate in the survey. Perhaps they did not consider sleep-inducing drugs an important issue or were afraid that this survey, although anonymous, might uncover a lack of pharmacological knowledge.

Conclusions and implications for practice
The personal trade-off between the perceived benefits and adverse effects of medicines is essential for deciding on their use [25]. Z-drugs seem to be attractive because experiential knowledge overemphasizes their benefits and fails to take risks such as drug-related falls and confusion into account. Moreover, difficulties to judge a drug's risk-benefit ratio do not prevent doctors and nurses from using them. In light of the dominance of experiential knowledge over descriptive knowledge, hospitals and clinical pharmacologists should not put too much faith in traditional continuous medical education about the risks of Zdrugs to reduce their usage. Rather, descriptive knowledge about Z-drugs should be accompanied at least by a numerical expression of risks and benefits, similar to a number needed to treat statistic. For example, Lee et al. explain that 13 older patients need to be treated with a benzodiazepine or Z-drug for one person to experience improvement in sleep quality, but only 6 patients need to be treated with these drugs for one person to experience an adverse advent [18]. Presenting nurses and doctors with a case series of patients showing these relative frequencies may have similar convincing power as experiential knowledge acquired during practice. In particular, results from local quality assurance data about relevant patient risks resulting from Z-drug use in one's own hospital (e.g., statistics about falls and craving) could be used as a case series for feedback and, thus, be a key to changing professional behavior.