Do older people know why they take benzodiazepines? A national French cross‐sectional survey of long‐term consumers

Abstract Objectives Benzodiazepines and non‐benzodiazepine hypnotics (or Z‐drugs) (BZD/Z) are widely prescribed for older patients despite major side effects and risks when chronically used. The patient's understanding of the treatment is one of the keys to good adherence. The purpose of the study was to assess the knowledge of BZD/Z treatment among older people who were taking BZD/Z for the long term by studying the concordance between the declared reason for taking BZD/Z and its indication. Methods This was a cross‐sectional, pharmacoepidemiologic ancillary of a national study. Data were collected through a semi‐structured interview. All patients from the main study were included. “Good knowledge” was considered when patients gave an indication for each BZD/Z that was similar to its marketing authorisation. Univariate and multivariate analyses were carried out to adequately determine profiles and characterise associations. Results More than half of the patients (61.6%) had a good knowledge regarding their treatment. The presence of a psychiatric disorder, a mean duration of BZD/Z use of less than 120 months, a desire to stop treatment, educational status and number and type of BZD/Z used were significantly associated (P < .05) with good knowledge. In the multivariate analysis, only a psychiatric disorder, educational status and taking at least one hypnotic drug were associated with good knowledge. Conclusions At the time of shared medical decision, it appears essential to improve the knowledge of the treatment by the patient. The rate of patients with good knowledge of their BZD/Z treatment remains low and even lower than what was previously found in the literature for other drug classes. In contrast to patients with good knowledge, these data highlight the characteristics of patients with poor knowledge of their BZD/Z treatment, which may allow populations at risk to be targeted and enable education measures to be strengthened.

of their BZD/Z treatment, which may allow populations at risk to be targeted and enable education measures to be strengthened. Despite their initial good efficacy for anxiety and insomnia, BZD/Z use carries major concerns about side effects such as anterograde amnesia, altered psychomotor functions, behavior, memory trouble, altered conscience state and dependence 3 when chronically used. These risks are increased by altered pharmacokinetic and pharmacodynamic parameters in older people. 4 A previous survey in 1996 shows that previous failure to stop taking the medication could explain 32% of continuous prescriptions of hypnotics. 5 Given the increase in prescriptions and the difficulty of discontinuing these medications, the French health authority in 2010 established recommendations to help practitioners stop BZD/Z use. 6 Beyond progressively decreasing doses and stopping, the recommendations emphasise that the presence of a psychiatric disorder, comorbid dependence and previous unsuccessful attempts to stop are factors that could lead to unsuccessful cessation of treatment.
In a 2014 study, Gérardin et al 3 found a rate of 44% of unsuccessful attempts to stop treatment in older patients. Among the factors that could explain why older people failed to stop, dependence in older people 7 and lack of knowledge about the treatment seemed to be major concerns.
Knowing the indication of prescribed drugs is essential and contributes to providing patients with a central position regarding their health care. With this information, patients can address their difficulties to the practitioner to choose the best therapeutic option. The concept of "shared decision making" 8 tends to be increasingly important. Treatment knowledge has been investigated in several studies with heterogeneous methodologies and results. [9][10][11][12] The literature does not include any study that directly examines specific knowledge of the indications of BZD/Z treatment among older people. The aim of this exploratory study was to assess the knowledge of the indications of BZD/Z in people aged 65 years or older who had been taking BZD/Z. Thus, we estimate the prevalence of patients with good knowledge of the indications of BZD/Z treatment(s).

| Study oversight
This study is an explanatory ancillary study of a national observational prospective study conducted by the French Addictovigilance Network (FAN) from March 2012 to December 2015. The FAN is Libertés). All participants provided written informed consent in accordance with the Declaration of Helsinki. The study is registered as NCT01920581.

| Patients
To be eligible, patients had to be 65 or older, be treated with benzodiazepines or Z-drugs in ambulatory care for at least 3 months (the maximal duration of an anxiolytic prescription in France) and give informed consent. Patients who were not fluent in French and/or with major cognitive impairments that prevented them from understanding the questions were excluded from the study.

| Study procedures
The patient's recruitment occurred where patients received their treatment in pharmacies, which necessarily included eligible patients.
All patients were then interviewed by phone by a trained interviewer without knowledge of the practitioner and without disturbing the patient-doctor relationship. The data included sociodemographic data, health problems (physical health problems and psychiatric issues), tobacco and alcohol con-

| Outcomes
The main objective was to estimate the prevalence of patients with good knowledge of the indications of BZD/Z treatment(s).
The secondary objective was to characterize patient profiles associated with good knowledge of the indications of BZD/Z treatment(s).

| RESULTS
In total, 1023 patients recruited by more than 250 pharmacies in France were included in this study ( Table 1).
The participants' ages ranged from 65 to 95 years, and threequarters were women. One-quarter of this population had a baccalaureate degree or higher, and most were retired. Forty percent lived alone, and they were divided between urban (45%) and rural populations (55% The 1023 patients took 1221 BZD/Z treatments (average: 1.2 per patient). The most frequently prescribed treatments are described in Table 1.
Bypassing a prescription, that is, using an alternative means to obtain the drug, was reported by 1 patient in 3, and the vast majority of patients had desire to stop treatment.
In the population with good knowledge, patients had significantly more reported psychiatric problems (P = .02), a maximum treatment duration more likely shorter than 120 months (P = .02), more frequent desire to stop BZD/Z (P = .01) and fewer than two treatments per BZD/Z (P < .001). Patients with good knowledge used significantly solely (P < .001) or combined hypnotic BZD (P < .001). The educational level differed significantly between the two groups (P < .01).
Dependence items were more prominent in the group with good knowledge, but the difference was not significant (Table 1). Table 2 shows the data for the multivariate analysis of the BZD/Z fit. In our model, three variables increased the probability of a good match between the treatment and the declared indication: declaring a psychiatric issue (P = .02), having a post bachelor's degree education (P < .001), a bachelor's degree or a technical diploma (P < .01) and taking only hypnotics (P < .001) or combined with anxiolytics (P < .001).

| DISCUSSION
One of the main contributions of this study is its description of the prevalence of good and bad knowledge about BZD/Z indications among older patients. This prevalence specifically in relation BZD/Z had never been clearly described before. The inability to report longterm treatment use to a health professional can be extremely deleterious or even dangerous. 13 Our study found that less than two-third of older patients were able to correctly report their reasons for taking BZD/Z, a rate that remains low.
A Chinese study 11  References. 15 Akici's study 12  The time gap between the doctor filling the prescription and the collection of data may have affected the lower knowledge rates. In contrast, at recruitment time, the patients were informed that they would be contacted specifically for a study on their BZD/Z medications and were asked to have their prescriptions with them at the call time, which suggests that the rates found may be overestimated.
Treatments such as hypnotics were more likely to be associated with good knowledge than anxiolytic. Not only did almost all the patients who took only hypnotic drugs have good knowledge, but the rates were even higher among patients who took both anxiolytics and hypnotics. These results are interesting because generally, the higher the number of treatments prescribed, the weaker the patient's knowledge about each treatment. 14,15,17,18 However, the multivariate analysis warrants a cautious interpretation of the odds ratios due to the small number of observations in some cases.
Insomnia is widespread (its prevalence in general practice is between 46% and 61% [19][20][21][22]  It is very surprising that only 11% of the patients reported a current psychiatric illness, although all of them had been using benzodiazepines for more than 3 months, which is already longer than the recommended prescription duration. Psychiatric disorders can be difficult for the patients themselves to recognize. This lack of insight may be explained by the pathology itself, by cognitive alterations in older people or by cultural bias. 23 One possible source of bias underlying underestimation is the choice to focus on current psychiatric issues, thereby failing to report essential underlying factors. For example, anxiety can be a symptom of psychiatric disorders as major depressive disorder, generalised anxiety disorder, etc.; meanwhile, BZD treat symptoms and not aetiologies. 6 It is possible that because the patient does not recognise and mention a psychiatric disease, only symptomatic treatment is prescribed, and information the more general underlying disorder is not noticed or integrated. In such cases, the result can be a very long mean duration of BZD/Z prescription (durations up to 10 years were observed in this study!), suggesting that prescription is continually renewed as the years go by, and BZD/Z are integrated as a "routine" of daily life without an understanding of the initial condition. This view is confirmed by the fact that mean prescription durations >120 months were more likely to be associated with poor knowledge. On the other hand, when a patient is able to recognise and self-report a psychiatric issue, it becomes clear that he is more likely to know why he or she is taking the medication.
Moreover, the desire to stop treatment was significantly more common in the good knowledge group. We can assume the existence of a kind of paradox in which patients do not know why they are taking the drug but do not want to stop it, and ultimately, that lack of knowledge makes discontinuing treatment more difficult.
The database of 1023 patients included in our multicentre study is particularly interesting and allowed a valid statistical analysis. Furthermore, it partly answers a crucial question that is known as a main factor in the successful cessation of treatment: do older people know why they take BZD/Z?
Comparing our older BZD/Z consumers to the general population, we found similar characteristics in terms of the sex ratio, educational level, prescribing practitioner and prescribed drugs, indicating that our sample was representative. 24 Our methodology was original compared to other studies. The fact that patients were recruited not at hospital discharge or from a GP's office but at their pharmacies and were contacted at their homes a few days later imparts a quality that is closer to the patient's "daily life" and is freed from evaluation bias.
Our study has several limitations. First, the definition of treat-

| CONCLUSION
Prescriptions of BZD/Z in older patients represent a major current challenge for doctors. This particularity is well illustrated in the difficulty of stopping BZD/Z use. In addition to dependence on the treatment, another possible reason for unsuccessful attempts to stop treatment is the patients' lack of knowledge about these medications.
These data highlight the characteristics of patients with poor knowledge of the drug, which may allow populations at risk to be targeted and enable education measures to be strengthened. This is particularly applicable for patients who take a treatment despite declaring no current health problems; although the treatment can achieve its objectives (anxiolysis and sedation), the aetiological problem does not disappear despite the improvement of physical, functional or paraclinical signs. These measures ultimately aim to improve adherence through improved drug knowledge.