Effect of the fixed combination of valerian, lemon balm, passionflower, and butterbur extracts (Ze 185) on the prescription pattern of benzodiazepines in hospitalized psychiatric patients—A retrospective case‐control investigation

Stress is an increasing problem that can result in various psychiatric and somatoform symptoms. Among others, benzodiazepines and valerian preparations are used to treat stress symptoms. The aim of this study was to investigate whether the prescription of a fixed herbal extract combination of valerian, lemon balm, passionflower, and butterbur (Ze 185) changes the prescription pattern of benzodiazepines in hospitalized psychiatric patients. In a retrospective case‐control study, anonymized medical record data from 3,252 psychiatric in‐house patients were analysed over a 3.5‐year period. Cases (n = 1,548) with a prescription of Ze 185 and controls (n = 1,704) were matched by age, gender, hospitalization interval, and main International Classification of Diseases, Version 10 F‐diagnoses. The primary objective was to investigate the effect of Ze 185 on the prescription pattern of benzodiazepines. Secondary objectives investigated the prescriptions of concomitant drugs and effectiveness of the hospital stay. Distribution of drug classes was analysed using the WHO's anatomic‐therapeutic‐chemical code. Data showed that both treatment modalities had a comparable clinical effectiveness but with significantly less prescriptions of benzodiazepines in the Ze 185 group (p = .006). This is of clinical importance because suitable alternatives to benzodiazepines are desirable. To obtain more support for this hypothesis, a dedicated randomized, controlled clinical trial monitoring drug safety is required.

respondents reported that they take medication to improve sleep. The most frequently taken sedative drugs were benzodiazepines as well as valerian preparations or benzodiazepine-like drugs (Tinguely, Landolt, & Cajochen, 2014). Long-term use of benzodiazepines can pose important adverse effects, which should be considered. These adverse effects include drug dependence, abuse, hangover effects, cognitive and memory impairment, drowsiness, ataxia, motor incoordination, and falls (Gerlach, Maust, Leong, Mavandadi, & Oslin, 2018;Johnson & Streltzer, 2013;Woods, Katz, & Winger, 1992). Especially in elderly, a more than 50% increased risk of hip fractures was found, not to mention consequential costs (Finkle et al., 2011;Johnson & Streltzer, 2013). As a consequence of important identified safety risks, an association between benzodiazepines and an increased risk of mortality was discussed (Agarwal & Landon, 2019;Donnelly, Bracchi, Hewitt, Routledge, & Carter, 2017;Palmaro, Dupouy, & Lapeyre-Mestre, 2015;Patorno, Glynn, Levin, Lee, & Huybrechts, 2017;Woods et al., 1992). Chronic use of benzodiazepines can lead to tolerance to the pharmacological effects and withdrawal symptoms after discontinuation of the drug. They act not only as sedatives but also have myorelaxant and anxiolytic effects. In addition, recreational use is a relevant complication (Lalive, Rudolph, Luscher, & Tan, 2011).
As stress causes various symptoms, such as anxiety, agitation, nervous tension, and sleep disorders, a single drug might be not the solution for the problem. A suitable way to treat stress-related symptoms may be the administration of a multicomponent mixture. Herbal  (EMA/HMPC, 2013, 2014. In Switzerland, complementary medicine, including treatment methods with evidence-based herbal medicine, is well accepted by patients and currently covered by the mandatory basic health insurance, when performed by a certified physician (Klein, Torchetti, Frei-Erb, & Wolf, 2015). There is one authorized herbal medicinal combination product (Ze 185) that, in addition to the above-mentioned herbal extracts, contains a dry extract of Petasites hybridus (L.) Gaertn., B. Mey. & Scherb (Asteraceae). P. hybridus is best known for its antispasmodic properties and anti-inflammatory effect (Anon., 2001). The efficacy and safety of Ze 185 for the treatment of somatoform disorders has been confirmed in a double-blind, randomized, placebo-controlled clinical study (Melzer, Schrader, Brattström, Schellenberg, & Saller, 2009). In addition, an open, randomized pilot study of the effect of Ze 185 in comparison with oxazepam was undertaken in patients with psychosomatic and psychovegetative disorders (Schellenberg, Sauer, & Brattström, 2004). Further, in a double-blind, randomized controlled clinical trial, the reaction to exam anxiety in healthy subjects was studied (Steiner & Opwis, 2000). Recently, the anxiolytic properties of Ze 185 were confirmed in a psychosocial stress paradigm: the Trier social stress test (Meier et al., 2018). These clinical studies show that Ze 185 is a well-tolerated herbal medicinal product in the respective indications.

| Ethical approval
According to Swiss law, at the time of the investigation, an ethical approval by the responsible local ethics committee of the Canton of Zurich was not necessary for the retrospective analysis of anonymized patient data (KEK, 2010).
The recommended dose is one film-coated tablet three times daily.

| Patient inclusion
Patients from all wards of the psychiatric clinic were included in the study. Therefore, cases and controls were selected for analysis by automatic identification from medical records using the Clinical Information System (CIS). Complete anonymization of the patient data was guaranteed for the analysis.
Cases: All patients were included in the analysis, if they were treated with any dose of Ze 185 on at least 1 day during their inpatient visit between January 2, 2010 and May 15, 2013. Controls: The control population without Ze 185 treatment was defined based on matching age (±5 years), gender, hospitalisation during the same interval (±6 months), and main International Classification of Diseases, Version 10 (ICD-10) F-diagnosis at the end of the clinical visit.

| Study variables
The primary objective of the retrospective data analysis was to investigate the effect of Ze 185 on the prescription pattern of benzodiazepines in hospitalized patients with psychiatric disorders. The prescription pattern was investigated stratified by various primary F-diagnoses according to ICD-10 (WHO, 1992) at the end of the hospital stay.
The secondary objectives were to investigate (a) the effect of Ze 185 on the prescription pattern of other concomitantly prescribed drugs and (b) the effectiveness of the hospital stay in comparison with patients without Ze 185 treatment. This effectiveness was evaluated based on the Clinical Global Impression (CGI) score (Guy, 1976), the Global Assessment of Functioning (GAF) score (Hall, 1995), and selective symptoms of the "Association for Methodology and Documentation in Psychiatry" (AMDP) system (AMDP, 1982;AMDP-CIPS, 1990).
In the CGI score, higher values correspond to a condition with stronger severity and lower values to less severity. In the GAF score, lower value counts correspond to a less severe disease and higher values to a stronger severity. In the AMDP system, 140 items are counted to provide a psychopathological diagnosis. The single items of the AMDP system were scored on a 4-point scale (0 = not at all, 1 = mild, 2 = moderate, and 3 = severe). For this retrospective data analysis, 11 items of the AMDP system were selected for detailed analysis. Four items from the category mental diagnosis (item 65 "anxious"; item 69 "restless"; item 82 "agitated"; item 83 "motor restlessness") and seven items from the category somatic diagnosis (item 101 "initial insomnia"; item 102 "middle insomnia"; item 106 "decreased appetite"; item 119 "palpitations"; item 122 "sweating"; item 126 "diffuse pressure in head"; item 127 "back pain") were chosen for further analysis.
The distribution of the drug classes was analysed using the WHO's anatomic-therapeutic-chemical classification system code with focus on medication with psychotropic activity.
As for safety, no data were reported in the CIS. Therefore, safety and tolerability of the patients' medication could not be evaluated.

| Database structure
For the descriptive data analysis, medical records from the CIS were extracted to an Excel file. The information scientist of the clinic extracted and anonymized all patient data. For the anonymization, the patient information number was used by the information scientist to create a pseudo patient information number. Thus, a complete anonymization of the patient data was guaranteed for the data analyst.

| Statistical analysis
Descriptive statistical analysis and all statistical tests used were performed using IBM SPSS Statistics for Windows, Version 21.0 (IBM Corp., Armonk, New York). The t test was used for normally distributed variables; otherwise, the Wilcoxon rank sum test was used. For the comparison of categorical variables, the Chi-squared test was used. The default summary statistics for quantitative and ordinal variables were the number of observations (n), mean, standard deviation (SD), or median, as appropriate. For qualitative variables, the number and frequency of subjects with non-missing data (n, %) per category were presented.

| RESULTS
A total of 3,252 patients were included in the retrospective data analysis. Of these, 1,548 cases were treated with Ze 185, and 1,704 matched controls received regular medical treatment without Ze 185. Demographic characteristics are described in Table 1. A distribution of the patients' ages at entry is shown in Figure 1. In both groups, age and gender were equally distributed.
The duration of the hospital stay was on average (mean ± SD)

| F-diagnoses
For the detailed analysis of F-diagnoses, focus was on the primary Fdiagnoses according to ICD-10 at the end of hospital stay ( Figure 2). In the main categories of ICD-10, the most frequent primary diagnosis was F3 "mood [affective] disorders" (cases n = 796; controls n = 866).
The most frequent subcategories were F32 "depressive episodes" (cases n = 319; controls n = 357) and F33 "recurrent depressive disorders" (cases n = 424; controls n = 449). The second most frequent primary diagnosis was F1 "mental and behavioural disorders due to psychoactive substance use" (cases n = 278; controls n = 275). These patients were mainly diagnosed with the subcategory F10 "use of alcohol" (cases n = 134; controls n = 165). These categories were equally distributed between the groups. The results showed a statistically significant difference between the groups for the primary diagnoses F2 "schizophrenia, schizotypal and delusional disorders" (cases n = 101; controls n = 181; p < .001) and F4 "neurotic, stress-related and somatoform disorders" (cases n = 185; controls n = 163; p = .028).

| Symptom severity and effectiveness
At baseline, controls showed a numerically comparable (6.07 ± 0.72 vs. 6.02 ± 0.67, respectively) but significantly (p < .022) higher CGI To evaluate the effectiveness of the hospital stay, the CGI score, GAF score, and selected items of the AMDP system were analysed.
At the end of therapy, the CGI score was similar in both groups (cases 3.36 ± 0.91, n = 1,421; controls 3.40 ± 0.95, n = 1,551; p = .289) and showed a significant improvement in comparison with the beginning of the therapy (p < .001) in both groups.
Importantly, in both groups, there was a significant improvement from baseline to the end of therapy (p < .001).

| Concomitant medication
Cases and controls received a plethora of concomitant medications.
Therefore, the population was analysed for differences between the groups. Based on the indication of Ze 185, special focus was put on medications with psychotropic activity. The numbers of prescriptions of anxiolytics (especially benzodiazepines), hypnotics/sedatives, and antidepressants were significantly different between cases and controls (Table 2).
Significantly less cases than controls received prescriptions of benzodiazepines (cases n = 661, controls n = 809; p = .006). The number of patients with prescribed hypnotics/sedatives was significantly higher for cases than controls (cases n = 940, controls n = 603; p < .001). In the group of hypnotics/sedatives, the main differences were seen for a specific valerian/hops extract (Ze 91019). Significantly more cases than controls received a prescription of this fixed extract combination (cases n = 784, controls n = 444; p < .001). The number of patients receiving antidepressant prescriptions was also significantly increased among cases compared with controls (cases n = 1,135, controls n = 1,114; p < .001). Of these, 496 cases and 439 controls received a prescription of selective serotonin reuptake inhibitors (p < .001). Significantly more cases than controls received prescriptions of other antidepressants (cases n = 778, controls n = 772; p = .005) including Hypericum extract (cases n = 50, controls n = 29; p = .005).

| DISCUSSION
The number of prescriptions of benzodiazepines has considerably increased over the past decades (Agarwal & Landon, 2019;Kaufmann, Spira, Alexander, Rutkow, & Mojtabai, 2016;Verthein, Buth, Holzbach, Neumann-Runde, & Martens, 2019). This development is giving cause for concern because of the increased risk of related adverse drug reactions posing a significant public health problem (Del Giorno, Ceschi, & Gabutti, 2017). It was concluded that understanding and addressing prescription patterns may help curb the growing use of benzodiazepines (Agarwal & Landon, 2019).
This retrospective case-control study investigated, among other aspects, the prescriptions and use of benzodiazepines in 3,252 psychiatric patients. Our data showed that both treatment modalities for cases and controls had a comparable clinical effectiveness as indicated by CGI, GAF, and AMDP scores. Cases received significantly less prescriptions of benzodiazepine (p = .006) but, on the other hand, more prescriptions of hypnotics (in particular valerian/ hops extract).
The demographic characteristics were similar in cases and controls. Due to the high number of patients in the case and control groups, there were small differences, which turned out to be statistically significant (e.g., for duration of hospital stay and baseline CGI score). However, these differences are probably clinically irrelevant.
T A B L E 2 Overview of prescribed medication in "cases" with Ze 185 and "controls" by ATC code in the drug class N, Nervous system Most patients had a primary F3-diagnosis such as F32 ("depressive episodes") and F33 ("recurrent depressive disorders"). Depression is often associated with anxiety and somatoform disorders. This is also reflected in the study population as the second most frequent diagnosis belonged to the category F4 "neurotic, stress-related and somatoform disorders". Ze 185 is effective in the treatment of depression and anxiety in patients with somatoform disorders (Melzer et al., 2009). As approved by the Swiss Agency for Therapeutic Products (Swissmedic), Ze 185 is indicated for the treatment of the following complaints: nervousness, tension, restlessness, and exam anxiety, leading to symptoms such as spasmodic gastrointestinal complaints, increased irritability, occasional trouble falling asleep, and sleeping through the night. Similar symptoms such as "reduced concentration," "disturbed sleep," and "somatic symptoms such as agitation" are also described in the F32-and F33-diagnoses. Therefore, the prescription pattern of Ze 185 reflected the authorized indication very well.
In addition to Ze 185, cases received prescriptions of other con-

| Strengths and limitations
Randomized, controlled clinical trials (RCT) are regarded as providing the highest evidence of proving treatment effects in evidence-based medicine. These studies possess a high internal validity as the investigator controls for all possible confounding effects. Observational studies (cross-sectional or case-control studies), however, have a higher external validity, because data are observed unbiased by protocol restrictions (Carlson & Morrison, 2009). As Vandenbroucke (Vandenbroucke, 2011) pointed out in an editorial, there are four meta-analyses contrasting RCTs and observational studies of treatments that found no large differences between the study types (Benson & Hartz, 2000;Concato, Shah, & Horwitz, 2000;Ioannidis et al., 2001;MacLehose et al., 2000).
The strength of this retrospective, case-control study is the avail- and AMDP items are due to missing data, which is a general limitation of retrospective studies (real world data) (Katkade, Sanders, & Zou, 2018). A multicentre study would have provided a broader picture.
Therefore, with regard to effectiveness, only hypotheses on the therapeutic setting can be generated. There might be a confounding bias due to the patients' and/or physicians' preferences regarding the uses of benzodiazepines or herbal drugs that influenced the prescription pattern. Neither a causal relationship between Ze 185 and the reduction of benzodiazepine prescriptions nor a better tolerability could be inferred, due to the CIS database lacking detailed adverse event monitoring.

| CONCLUSION
The broad indication of Ze 185 covers the treatment of stress-related complaints, such as nervousness, nervous tension, agitation, and anxiousness. Patients with these symptoms are often treated with benzodiazepines. The data in the present retrospective case-control study have certain limitations but provide some evidence that a treatment modality including Ze 185 could reduce the need for benzodiazepines. This is of general clinical interest and relevance as benzodiazepines are under ongoing discussion due to their problematic safety profile. However, to obtain a solid answer for this hypothesis, a dedicated randomized, controlled clinical trial closely monitoring drug safety is needed.

ACKNOWLEDGMENT
This study was conducted as investigator-initiated research. There was no funding by the manufacturers of the investigated herbal medicinal products or other mentioned central nervous system drugs.

CONFLICT OF INTEREST
MK and KS were employees of the Clienia Private Clinic Schlössli. SN, CB, CZ, and JD are/were employees of Max Zeller Söhne AG and supported the analysis of the data and the preparation of manuscript on the basis of mutual scientific agreement with MK and KS. As such, the employees confirm no bias to the publication.

AUTHOR CONTRIBUTIONS
MK was involved in supervision, study design, data interpretation, and manuscript preparation. KS contributed to data collection and analysis. SN and CB were involved in the manuscript preparation and data interpretation. JD and CZ contributed to study design, data analysis, manuscript preparation, and revision.