Efficacy and safety of lavender essential oil (Silexan) capsules among patients suffering from anxiety disorders: A network meta-analysis

A systematic review and network-meta analysis (NMA) were performed to estimate significance of the anxiolytic effect of lavender essential oil taken as silexan capsules versus other comparators (i.e., placebo/paroxetine/lorazepam). The outcome of interest was Hamilton Anxiety Scale (HAMA). Weighted mean differences (WMD) were calculated to estimate the treatment effect at the confidence interval of 95%. League tables were generated using treatment effect, for all pairwise comparisons, where WMD < 0 favors the column-defining treatment. Five studies were identified with a total of 524 participants receiving treatment with silexan 80 mg and 121 participants taking silexan 160 mg. The NMA results indicated that consumption of silexan 160 mg resulted in higher decline of HAMA score [WMD −1.14 (−1.10, 3.39)] in comparison to silexan 80 mg, placebo [−2.20 (−4.64, 0.24)] and paroxetine [−1.24 (−5.34, 2.85)]. The effect of silexan 80 mg was observed to be same as that of paroxetine. Overall, silexan 160 mg was noticed to be a more efficient treatment giving significant decline in HAMA score across other comparators. However, no improvements in HAMA score was observed for the group receiving lorazepam 0.5 mg when compared to silexan 160 mg, silexan 80 mg, paroxetine 20 mg, and placebo.

with anxiety disorders. However, the chronic character of anxiety disorders and requirements of the long-term treatment set very high safety and compliance standards for the medications and bring phytotherapy as a treatment option 12 .
Phytotherapy has been gaining popularity in the treatment of anxiety [13][14][15][16][17] with many GABA-modulating medicines of herbal origin undergoing preclinical and clinical investigations 18 . Particular attention has been paid to anxiolytic-like effects of essential oils, among which a lavender essential oil demonstrated the best pharmacological profile [19][20][21] .
Lavender (Lavandula angustifolia Miller or Lavandula officinalis Chaix) has a long history of traditional use and its essential oil was found to possess a wide range of biological effects 22,23 . Evidence of the effectiveness of the lavender essential oil in the pharmacotherapy of mental disorders led to the development of Silexan, which is a standardized essential oil of L. angustifolia flowers prepared by steam distillation 24 . Silexan is approved in Germany for the treatment of restlessness related to anxiety and marketed as LASEA ® 25 .
Silexan was found to contain 36.8% of linalool and 34.2% of linalyl acetate 26 . Other components of the lavender essential oil present in substantial quantities include monoterpene alcohol lavandulol, its ester lavandulyl acetate, and bicyclic monoterpenoids borneol, eucalyptol (1,8-cineole) and camphor 27,28 . Linalool also demonstrated anxiolytic properties in several animal models 29,30 and possessed stress-relieving effect in humans under the experimental stress 31 . Studies on molecular mechanisms of pharmacological effects of lavender essential oil revealed that its effect on CNS could be attributed to the inhibition of voltage dependent calcium channels 26 . Unlike some other monoterpenes found in the lavender essential oil, linalool and linalyl acetate significantly inhibited voltage dependent calcium channels 26 . In another study, lavender essential oil was demonstrated to possess affinity to the NMDA receptor and SERT 28 . Binding to NMDA receptor was also observed for major constituents of lavender essential oil linalyl acetate and linalool, while only linalool demonstrated significant binding to SERT 28 . Further results of the clinical, randomized, placebo-controlled, double-blind, cross-over study on healthy men using positron emission tomography technology suggested that the anxiolytic effect of lavender essential oil (given as Silexan) could be attributed to the serotonergic system changes, particularly at the 5-HT 1A receptor level 32 .
To date, two reviews have been reported in the literature evaluating the efficacy of Silexan in anxietydisorders 33,34 . Review by Kasper 33 in 2013 was not a systematic review, liable to bias, and the author did not combine results of individual studies using meta-analysis. The systematic review and meta-analysis by Möller et al. 34 estimated the efficacy of Silexan in sub-threshold anxiety disorders and reported significant reduction in anxiety as measured by HAMA scale. The present systematic review further builds on our existing knowledge on the efficacy of Silexan in the management of anxiety disorders by comparing the efficacy of different strengths of lavender oil using a network meta-analysis technique. This is a unique technique and is considered superior to meta-analysis when direct multiple comparisons between the intervention and control group do not exist in the literature 35 . In addition, our systematic review has evaluated safety of lavender oil as well.

Methods
A systematic review was performed to identify potential research papers across 8 databases from inception till December 14, 2017. The NMA was performed, in accordance with the recommendations made by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 36 , to estimate the overall effectiveness of lavender versus placebo and other medicinal products in anxiety disorders and its adverse effects.

Population intervention comparator and outcomes.
• Population of interest: Patients with anxiety disorder.
• Intervention: Lavender given orally (capsules). All the relevant papers were identified and imported to an EndNote file to create a combined library. Duplicate records were removed, where appropriate. Title and abstracts were screened by two authors independently and disagreements were resolved through discussions. Full-texts were downloaded and considered for inclusion based on the inclusion/exclusion criteria described below. Also, we manually reviewed the references of the included studies to identify other relevant studies. Additional details about the studies identified are shown in Fig. 1.

Study selection.
The following criteria were used to further align the search with the specific outcomes of interest.

Inclusion criteria.
• All English language studies from inception till December 14, 2017 were considered eligible for inclusion in this systematic review.
• Only human experimental studies, which are clinical trials comparing the effect of lavender (as a tea, powder or capsule) were considered for potential inclusion in order to assess causal assertions. Comparators considered appropriate were: a placebo, reference/control group, or any other active regimen, which was compared with the lavender formulation given orally. • The main outcome of interest is the change in Hamilton Anxiety Scale (HAMA) total score of the patients.

Exclusion criteria.
• Systematic reviews, all observational studies, letter to the editors, case reports, case series, personal opinion, qualitative studies, and reviews/communications focusing on individual experience of the use of lavender were excluded. • In addition, any data that are not published, reports, and thesis were excluded.
• Experimental studies on animals and studies using lavender for aromatherapy were excluded.

Outcome of interest.
• Primary outcomes: Change in Hamilton Anxiety Scale (HAMA) total score of the patients. • Secondary outcome: Safety and tolerability of lavender preparations as a medicine.
Data extraction. Data were extracted independently by two review authors (WSY, TMK) and in the case there was a conflict, a third independent assessor AVD assisted in revalidating the data and resolving the conflict.
A structured, pilot-tested data collection form was designed to collect data from individual studies. In addition to the data related to the outcomes of interest, data on the number of authors, year of publication, study design, setting and country where the study took place, sample size, patients' mean age and gender, and inclusion criteria and exclusion criteria were also extracted. The results for the outcome measures included in this review were summarized as mean and SD difference from baseline to endpoint in both the intervention and control groups. When not reported, the mean and SD difference were calculated provided sufficient data were available. risk of bias assessment tool 37 . Two authors (MAH, ZJ) assessed risk of bias independently and disagreements were resolved through discussion.
The network meta-analysis (NMA) was performed using STATA version 14 ® . Random effects model was used and weighted mean differences with 95% confidence interval (CI) were calculated for all continuous outcome measures to compare the effectiveness of the intervention. P-value less than 0.05 was considered statistically significant. Furthermore, in order to generate forest plots for NMA, respective pairwise comparisons for the treatment effect were carried. League tables were created using treatment effect calculated based on weighted mean difference (MD; 95% CI) for all pairwise comparisons. A league table, a square matrix consisting of all pairwise comparison within a meta-analysis, empowers researcher/reader to directly compare the direction and magnitude of treatment effect, encouraging easier interpretation of results 35 . A negative (−) MD indicate decline in the score for the HAMA and SAS scales. www.nature.com/scientificreports www.nature.com/scientificreports/

Results
Included studies. Of the 982 studies screened, five randomized control trials were included in this review ( Fig. 1).

Characteristics of the studies and type of interventions.
To assess the effect of lavender in treating anxiety, interventions included in the 5 studies were placebo, 80 mg/d silexan, 160 mg/d silexan, 0.5 mg/d lorazepam, and 20 mg/d paroxetine. The study by Woelk & Schalfke 38 had a treatment phase of 6 weeks whereas the other 4 studies by Kasper et al. [39][40][41][42] had a treatment phase of 10 weeks. All these studies were conducted in Germany. The primary outcome in each study was the change in total score of HAMA. Safety and tolerability of the interventions were assessed by the reporting of adverse events that occurred over the course of the study. Details on study characteristics are available in Table 1.
Although the recruitment criteria for all 5 studies included participants of both genders and of any ethnicity, less than 31% of participants in each study were male and almost all were Caucasian. Woelk    www.nature.com/scientificreports www.nature.com/scientificreports/ conducted a multi-center, double-blind, randomised phase III study assessing the comparative effect of 80 mg silexan daily versus 0.5 mg lorazepam daily on a sample size of 78 participants with 23.4% being male. In this study, 40 participants received silexan 80 mg. Kasper et al. 39 investigated the effect of 80 mg silexan daily against placebo in a randomized, double-blind, multi-center trial with 20.2% of its 228 participants being male. In this study, 104 participants received treatment with silexan 80 mg.
Following that, Kasper et al. 40 conducted a randomized, double-blind, double-dummy, multi-centre trial with four parallel groups receiving 160 mg silexan daily (121 patients), 80 mg silexan daily (135 patients), 20 mg paroxetine daily, and placebo with 26.4% of male in the sample size of 616 participants. Subsequently, Kasper et al. in 2015 41 investigated the effect of 80 mg silexan daily and used placebo for the reference group. The study design was a randomized, double-blind trial with 28.2% of the study's 179 participants being male 38 . In this study, 86 participants were assigned to the silexan 80 mg group. The following year, Kasper et al. published another paper to again investigate the effect of 80 mg silexan daily compared to placebo 42 . In this double-blind, randomized,  www.nature.com/scientificreports www.nature.com/scientificreports/ parallel-group multi-centre trial, the sample size was 362 participants and 30.8% of them were male. The silexan 80 mg group consisted of 159 participants 42 . Further details on participant characteristics are shown in Table 2.

Risk of bias.
In general, the overall risk of bias was low for most of the domains for the included studies ( Figs. 2 and 3). In particular, risk was low for selection bias, performance bias and detection bias for all the studies. One of the studies has unclear risk of bias for allocation concealment (Figs. 2 and 3). In other biases, studies were deemed to be high risk of bias as they were funded by the manufacturer.
Effect on HAMA score: Primary outcome. To estimate the effect of all interventions on HAMA, five studies were included and a NMA was performed to compare and contrast the effect on the HAMA scale. The network plots demonstrate the organization of all available evidences. The width of the lines represents the number of trials and the size of node represents the sample size (Fig. 4).  Table 3.

Overall analysis. Outcome of interventions versus placebo at
Furthermore, a pairwise analysis was carried out to estimate the comparative effect of all the treatments versus placebo and other interventions. Using the pairwise comparison, a league table was generated to present all possible pairwise comparisons between any two of the five treatments (Fig. 5). Treatment effect, that is, the weighted mean difference of each pairwise comparison was calculated and reported along with the 95% CI. It was revealed that lorazepam 0.5 mg had no positive effect and did not cause a decline in HAMA score. The treatment with silexan 160 mg was found to be effective in comparison to placebo, paroxetine 20 mg, and lorazepam 0.5 mg. Overall, the effect of silexan 160 mg [−1.14 (−1.10, 3.39)], assessed on the basis of declining HAMA score, was superior in comparison to silexan 80 mg. In addition, the effect of silexan 80 mg [−1.06 (−3.32,1.21)] was observed to be slightly better than that of placebo. The decline in HAMA score was higher for silexan 160 mg in comparison to placebo and paroxetine 20 mg. Overall, all interventions including placebo were observed to be more effective in declining HAMA score compared to lorazepam 0.5 mg. Details are described in Table 4.

Safety of silexan.
In the analyzed 5 studies, adverse events, attributable to silexan use, consisted mainly of gastrointestinal problems such as nausea, eructation or breath odour, and diarrhea. A number of patients also reported having headaches in the 2015 study by Kasper et al. 41 Nevertheless, the number of patients that experienced these mild adverse events comprised of a small percentage of the sample size. No serious adverse event was found to be linked to the use of silexan.  Table 3. Weighted mean difference of interventions versus placebo. I 2 = 0.00%, Tau 2 = 0. www.nature.com/scientificreports www.nature.com/scientificreports/ Fatigue is a known side effect of lorazepam. In the study conducted by Woelk & Schlafke 38 , 16.2% of patients receiving treatment with lorazepam had fatigue whereas none in the silexan group experienced fatigue. However, 10.0% in the silexan group had nausea whereas only 2.7% had nausea in the lorazepam group.
The actual number and type of total adverse events experienced by the participants in the study conducted by Kasper et al. 36,37 were not reported in details. Nonetheless, it was noted by Kasper et al. 37 that the participants receiving treatment with silexan experienced a 3% increase in risk of gastrointestinal problems compared to those in the placebo group. Further details on adverse events reported in the studies can be found in Table 5.

Discussion
To date, perhaps it is the first NMA estimating the effect of oral dosage form of lavender i.e. silexan in comparison to placebo, paroxetine and lorazepam on the anxiety score at the HAMA scale. Systematic reviews and meta-analysis have proven to be an effective tool to estimate the clinical efficacy and safety of herbs, and further comparative analysis using network meta-analysis will provide researcher an opportunity to estimate the clinical efficacy and safety of a herb versus various treatments or comparators 43 . Therefore, the current systematic review and NMA will serve as a useful tool in analyzing the use of silexan in clinical settings in an effective manner. The overall analysis revealed that administration of silexan 160 mg produced a higher decline in HAMA score [WMD −1.14 (−1.10, 3.39)] in comparison to silexan 80 mg. The effect of silexan 80 mg was observed to be similar to paroxetine 20 mg. Overall, the treatment with silexan 160 mg resulted in more significant decline in the HAMA score across all the comparators and therefore can be considered as the most effective anxiolytic intervention for the included studies. However, it should be noted that the dose of lorazepam used in the included RCT, and subsequently in this NMA, was only 0.5 mg which is the lowest daily dose recommended. Comparative efficacy of silexan with higher dose of lorazepam is not known and needs further research.
The lavender in different forms has been used in cosmetic and therapeutic applications for centuries. These applications were driven by unique scent of lavender essential oil. This essential oil became popular in aromatherapy and its therapeutic effectiveness in this form has been assessed clinically in a number of trials. Particularly, aromatherapy with lavender essential oil was effective in managing preoperative stress and anxiety in various settings significantly decreasing anxiety level compared to placebo [44][45][46] . However, similar interventions applied after surgery revealed no difference with control group on anxiety level and mental stress 47,48 .
The aromatherapy with lavender essential oil improved quality of sleep and reduced level of anxiety in patients with coronary artery disease 49 , significantly reduced anxiety in older patients with acute coronary syndrome 50 and myocardial infarction 51 . Lower indexes for perceived stress and objective stress were observed in the intensive care unit patients receiving aromatherapy with lavender essential oil 52 . The inhaled lavender essential oil helped to alleviate anxiety in burn patients 53 and postpartum women 54 .
However, no significant long-term improvements were observed in control of anxiety for patients suffering from cancer receiving aromatherapy treatment with lavender essential oil 55 . No significant difference in the anxiety level was also observed between cancer patients receiving inhalation with lavender essential oil during radiotherapy and the placebo group 56 . Overall, the aromatherapy with lavender as an anxiolytic agent has not been supported by sufficient evidence of therapeutic efficacy 57 . However, our NMA was based on the systemic administration of the lavender essential oil (Silexan) and demonstrated positive evidences for its anxiolytic effect.   Table 5. Side-effects reported among the participant consumed placebo and intervention treatment.

Strengths and limitations. The exclusion of non-English articles can be one of the limitations of this
NMA. This was mainly due to a lack of resources to translate articles written in other languages. There is a slight possibility that some relevant data from these non-English literature, if combined with the current analysis, might change the significance of lavender effects versus other comparators. In addition, all the studies investigating the effect of lavender used its essential oil in form of Silexan capsules. These studies were from Germany and the outcome might be potentially affected by the patients' genetics. To get more comprehensive analysis and conclusions on the effect of lavender and particularly Silexan capsules, similar clinical studies using ethnically more diverse populations are required. The lavender preparations like Silexan capsules might have different results on HAMA scale in other races from different genetic makeup. Moreover, the current analysis estimated only the overall effect of the treatments on HAMA score, the current study was unable to compare the multiple treatments based on the decline in symptoms, severity and relapse of anxiety, which was mainly due to variable reporting of the results across the selected studied for the NMA. Overall, the other bias was observed across all the five included studies. However, one of the main strength of this NMA is heterogeneity among the one-on-one comparison and pairwise comparison was 0% which reflects no inconsistencies among the studies. A number of factors including similarities between the study populations across RCTs included in this NMA as all studies were conducted in Germany, similarities in intervention (all in the form of Silexan capsules) and duration of follow-up and use of same tool (HAMA) to assess anxiety in all included studies could explain low heterogeneity observed in NMA.

Conclusion
Results (through pairwise comparisons) revealed statistically significant effect of silexan 160 mg versus placebo, silexan 80 mg and paroxetine 20 mg. In addition, the effect of paroxetine 20 mg was also observed superior than placebo and silexan 80 mg. However, silexan consumption has shown some gastrointestinal side effect such as nausea, eructation or breath odor and diarrhea, which were tolerated by the patients recruited in the silexan arm. There is a need of more methodological strong studies to further investigate the effect of silexan among the patients from other regions to get a comprehensive picture about its clinical efficacy and safety.

Data availability
All materials, data and associated protocols are available to readers without restrictions.