In our prospective, observational study we found a high prevalence of chronic hypnotic drug use, i.e. 26.6% of all patients admitted to the geriatric ward. We included 96 patients using a BZD or a Z-drug chronically. Hypnotic drug use was discontinued in 36.5% of the patients at discharge. A small minority of the patients (4.2%) discontinued but was relapsed at discharge, while many patients (26.0%) were relapsed one month after discharge. The main determinant for drug discontinuation at one month was deprescribing to have occurred during hospital stay. Of all participants, 21.9% successfully discontinued hypnotic drug use one month after discharge.
Previously reported estimates concerning long-term hypnotic drug use vary from 20% to more than 50%, depending on the age and residence of the study population, but this number has been decreasing over the past years [20, 21, 30–32]. The prevalence in our population (26.6%) is therefore rather low. Indeed, the true number of long-term hypnotic drug users might be an underestimation as patients who used multiple hypnotic drugs or had already discontinued were not considered for study participation.
Our findings on discontinuation were higher than owing to voluntary cessation in outpatient care due to the deprescribing culture among the physicians and awareness campaigns in the hospital. In a recent retrospective study performed in the United States, spontaneous discontinuation of chronic BZD use in adults aged 18 years and older was only 13.4% at the end of the following year which is much lower than in our study [33]. The discontinuation rate of the control group in the EMPOWER trial was 5% [34]. A patient empowerment intervention in combination with a personalized tapering protocol, on the other hand, resulted in a BZD discontinuation rate of 27% demonstrating the added value of such an approach. However, this trial was performed in community-dwelling adults aged 65 and older who received little support, which might have mitigated the impact of the intervention [34].
Since the median LOS of the participants was 12 days, discontinuation was performed within a limited time window, indicating that short-term withdrawal might be feasible in the hospital setting. Our results showed that discontinuation of the hypnotic drug during hospitalization was strongly associated with discontinuation one month after discharge. This can be explained by the fact that discontinuation during hospitalization occurred under close monitoring of a physician and that patients were only discharged when they were considered to be clinically stable. Importantly, our results also showed that there was no difference in sleep quality between the two groups or within groups over time, demonstrating that hypnotic drug discontinuation was not associated with deterioration of sleep quality. In fact, overall sleep quality in the study population was poor, reflected by a median total PSQI score higher than 5 at all follow-up moments [29]. It was remarkable that two variables related to experiencing a delirium during hospitalization were significant in the UVA (i.e. abrupt discontinuation during hospitalization, delirium during hospitalization) but not in the MVA. These variables were expected to negatively contribute to hypnotic discontinuation one month after discharge. A possible explanation for this could be that insufficient participants experienced a delirium to significantly contribute to the hypnotic discontinuation rate.
A first major strength of our study was that we enrolled geriatric patients, including very old subjects. To our knowledge, this was the first study of this magnitude investigating hypnotic discontinuation in this subset of the geriatric population. Second, our definition of chronic hypnotic drug use was relatively short compared to previous reports [25, 35–37]. We deliberately chose this definition because guidelines specifically recommend to limit BZD or Z-drug use to maximally four weeks [38]. Third, the median MMSE of 55 out of the 96 participants was 24, which is considered to be the lower limit of normal cognition [39]. We were hence able to include a larger number of cognitively impaired older adults than in previous investigations, where cognitive impairment was frequently defined as an exclusion criterion [25].
Our study had some limitations as well. First, no blood or urine sampling was performed to establish actual discontinuation. After discharge, we relied on information provided by the participants or their relatives. Second, even though the PSQI is a validated questionnaire, it could be questioned whether it gives an accurate measurement of sleep quality. Polysomnographic sleep measurement would have been a more objective measure, but this was not feasible and might have impaired the willingness to participate [29]. Third, the Hawthorne effect may have played a role as well since participating to a study may influence the reported sleep quality by the participants [40]. Fourth, despite clearly stating in plain language that this was an observational study, a large number of patients were still anxious about their hypnotic drug being discontinued. There was also a substantial influence of family members who were convinced that participating in this study would be an additional burden for already fragile patients. We hypothesize that this is one the major reasons why about one third of the patients eligible for enrollment refused to participate. Fifth, physicians already discontinued hypnotic drug use in 8.9% of the patients before the patient could be enrolled. In addition, when hypnotic drug use was not registered in the electronic record or when hypnotic discontinuation was performed at the emergency department, this was not taken into account. As a result, the true discontinuation rate might have been higher. Sixth, we attempted as much as possible to obtain information from the electronic record of the hospital or through other health-care professionals for non-hospitalized patients. Nevertheless, when patients were discharged home, we had to rely on information provided by the patient or his family, thereby possibly compromising the accuracy of the information. In addition, some information might not have been reliably documented in the electronic record such as whether or not the patient experienced a delirium during hospitalization. Seventh, some relevant variables were not documented and their contribution to hypnotic drug discontinuation could therefore not be investigated. For instance, an interesting variable would be the duration of hypnotic drug use before admission. However, we were unable to determine this variable with the required accuracy. Finally, we were unable to judge the long-term discontinuation rate (e.g. over 6 or 12 months), owing to the restrictions imposed by the Ethics Committee. They argued that the study might interfere too much in the relationship between patients and the primary care physicians. Nevertheless, the highest discontinuation rate was achieved three months after sending of a discontinuation letter [41]. About half of patients still relapsed thereafter, emphasizing the importance of a longer follow-up period.
An intervention to improve the discontinuation rate should focus on different aspects. Based on previous report, it might be beneficial to send a discontinuation letter to the patients’ GP and family in addition to deprescribing hypnotics drugs during hospital stay [41]. In fact, the individual healthcare professional could be the most important contributing factor to hypnotic discontinuation underlining the importance of involving the patients’ GP early on in the deprescribing process [33]. As the patient’s family had a considerable influence on the decision to participate in this study, their importance should not be overestimated. They should be informed about the disadvantages of chronic hypnotic drug use and should be involved to encourage the patient when necessary. Because recommendations of ward-based clinical pharmacists are generally well-accepted by physicians [42], formulating a well-considered advice regarding hypnotic drug use to the geriatrician is expected to have an impact as well. All together, these findings illustrate that a multifaceted intervention should be implemented to maximize the success rate of hypnotic discontinuation one month after discharge, while focusing on discontinuation during hospital stay.
In the post-implementation cohort of this study we will therefore implement an intervention consisting of four components. First, a standardized withdrawal scheme will be implemented, based on the protocol of Petrovic et al. [25]. Second, educational and informational moments will be offered to the ward-based physicians and nursing staff. Third, patients will be informed about the side effects of hypnotic drugs and the possibility of worsening sleep in the first days after discontinuation of the hypnotic drug. Fourth, the patient’s GP will be informed by phone and with a letter about the patient’s study participation. For patients going to a nursing home, the nurses will be informed by phone as well.