Mirtazapine for chronic insomnia in older adults: a randomised double-blind placebo-controlled trial—the MIRAGE study

Abstract

Background

Mirtazapine promotes sleep by blocking serotonin and histaminergic receptors and is often used off-label to treat chronic insomnia. However, its efficacy remains to be demonstrated in a clinical trial. The MIRAGE study aims to determine the efficacy and safety of mirtazapine in older patients with chronic insomnia.

Methods

This was a double-blind, randomised, placebo-controlled trial in a geriatric outpatient clinic of a teaching hospital. Adults aged 65 years and older with chronic insomnia were included. Sixty participants were randomised in a 1:1 ratio to receive mirtazapine 7.5 mg or a matching placebo for 28 days. The primary efficacy endpoint was the mean change in the Insomnia Severity Index (ISI) score from baseline to 28 days post-treatment. The primary safety endpoints included any adverse events reported during the clinical trial and all adverse events leading to premature discontinuation.

Results

Mirtazapine was superior to placebo on the primary outcome measure, subjective wake after sleep onset, total sleep time and sleep efficiency. After 28 days, the mean change in ISI score was significantly greater in the mirtazapine group (−6.5 [95%CI; −8.3 to −4.8]) compared to the placebo group (−2.9 [95%CI; −4.4 to −1.4]), with a p-value of 0.003. No participant experienced severe adverse events. A total of 6 participants in the mirtazapine group and 1 participant in the placebo group discontinued their treatment due to adverse events.

Conclusion

Mirtazapine reduces chronic insomnia symptoms in older people. However, its use may be limited by mild but clinically relevant adverse events. (clinicaltrials.gov NCT05247697).

Impact statement

This is the first randomised, double-blind, placebo-controlled trial on the efficacy and safety of Mirtazapine in older adults with chronic insomnia. Our findings show that a 28-day treatment with mirtazapine, compared to placebo, significantly reduces insomnia severity, as measured by the Insomnia Severity Index. Despite the current lack of robust evidence, mirtazapine is widely prescribed by clinicians to treat insomnia in older adults with chronic insomnia. Publishing our study will facilitate the broad dissemination of this critical information, helping clinicians more effectively treat their older patients.

Key Points

• Mirtazapine significantly reduces insomnia symptoms in older adults with chronic insomnia compared to the placebo.

• Mirtazapine caused more mild adverse events and treatment discontinuations than placebo.

• Mirtazapine’s acceptability and adherence may be hindered by its safety profile.

Introduction

In older adults, chronic insomnia causes significant sleep impairment combined with daytime symptoms such as lack of attention and sleepiness [1, 2]. It is associated with depressive symptoms, reduced quality of life and is a significant predictor of hypnotic prescriptions for insomnia [3, 4]. Cognitive-behavioural therapy for insomnia (CBTi) is the first line treatment for chronic insomnia in older patients [5, 6]. This therapy is effective long-term but requires many sessions, significant commitment from older adults and its benefits take time to appear. [7, 8] Financial constraints frequently limit access to psychotherapy [9] and older individuals are often hesitant to seek mental health support, leading to underutilisation of services. For these reasons, hypnotic drugs are also frequently prescribed. In older adults, guidelines recommend the use of benzodiazepines, Z-hypnotics and doxepin for no longer than 4 weeks, whilst orexin receptor antagonists (ORAs) may generally be used for up to 3 months [5, 6].

Long term use of benzodiazepines and Z-hypnotics lead to dependence [10, 11]. These medications reduce balance, increase the risk of accidental falls and cognitive decline and elevate mortality [12–14]. Despite these safety concerns, they remain the most prescribed treatment for insomnia [15–17]. Antidepressants such as trazodone and mirtazapine are also commonly used [16, 18, 19]. The 2023 European guideline suggest that antidepressants including mirtazapine could be used on a case by case basis with a discussion on risk and benefits [6]. Mirtazapine, a tetracyclic antidepressant, promotes sleep by blocking serotonin 2A and histaminergic H₁ receptors. Mirtazapine was associated to improvements in sleep efficiency, total sleep time and sleep quality in patients with depression, [20] which may have contributed to its use in treating insomnia [18, 19, 21]. However, there is a complete lack of substantial evidence from high-quality clinical trials to support its use for the treatment of insomnia. Only three clinical trials have investigated the efficacy of mirtazapine for sleep complaints, and neither was conducted in older participants with a formal diagnosis of insomnia according to consensus diagnostic criteria [22–24]. Mirtazapine has not been associated with dependence, and its link to falls and factures remains controversial [13, 25]. Given that mirtazapine may have a better safety profile than commonly used hypnotics, especially in older adults, it is of utmost importance to rigorously assess its efficacy and safety in a placebo-controlled clinical trial.

The MIRAGE study is the first randomised controlled trial assessing the efficacy and safety of mirtazapine in patients with chronic insomnia. It focuses specifically on older adults.

Methods

Study design

We conducted a randomised, double-blind, parallel group, controlled trial from October 2022 to July 2024 at the Centre Hospitalier de l’Université de Montréal (CHUM) in Canada. The study was approved by the CHUM’s ethics board, and all participants provided written informed consent. The results were reported in accordance with the Consolidated Standards of Reporting Trials (CONSORT) guidelines. (NCT05247697).

Participants

Participants were enrolled by the investigators at a booth in the hallway of the hospital’s outpatient clinics and from the community. Sleep disturbances were not the primary reason patients were referred to the clinics. Community participants were recruited through presentations, ads in pharmacies and a magazine for older adults. Eligible participants were adults age ≥65 years with chronic insomnia, as defined by the International Classification of Sleep disorders 3rd edition [26]. Participants had to report difficulty initiating or maintaining sleep, along with daytime complaints, occurring at least three times per week for 3 months. We excluded participants who met any of the following criteria: current use of any hypnotic drug, psychostimulant drugs or melatonin and current participation in CBTi. Detailed exclusion and inclusion criteria are listed in supplementary material.

Randomisation

Participants were randomly assigned in a 1:1 ratio to receive either mirtazapine 7.5 mg or a placebo once daily at bedtime. Previous Mirtazapine doses used for insomnia ranged from 7.5 to 30 mg [22, 27, 28]. The lowest dose was chosen since older adults have increased sensitivity to adverse reactions [29]. The research pharmacy generated a block-of-four randomisation list. Participants, investigators, hospital personnel and the sponsor were blinded to treatment allocation. The investigational treatment and matching placebo were indistinguishable and packaged identically. Unblinding during participation was allowed only in a medical emergency, but none occurred. After participation, unblinding was permitted once all participants in the randomisation block completed the trial, which happened once.

Study procedures

Chronic insomnia was diagnosed by a trained geriatrician. During the first week, participants did not take any study medication, and the data collected served as the baseline. For the following 4 weeks, participants took the research drug once daily at bedtime. After discontinuation of the research medication, data were collected for 7 days to assess the withdrawal effects of mirtazapine. During the 6-week clinical trial period, sleep diary and actigraphy data were collected during the first, fifth and sixth week. The Insomnia Severity Index (ISI) and Pittsburgh Sleep Quality Index (PSQI) questionnaire were administered on day 0 and day 36, that is, after 28 days of experimental treatment. Upon entering the study, all participants received a printed sheet outlining sleep hygiene measures [30]. The recommended measures focused on sleep habits and the sleep environment. Cognitive-behavioural therapy was not allowed during the course of the clinical trial. A study flow diagram is illustrated in Figure 1.

Assessment

Data obtained during the baseline interview by the principal investigator included demographic information, medical history, current medications and depressive symptoms (using the Montgomery-Asberg Depression Rating Scale, [MADRS]) [31]. Both sex (assigned at birth as female or male) and gender (identified as woman or man) were recorded. Exercise habits were categorised as "never" (no exercise), "occasional" (less than once a week) and "regular" (once a week or more). Cognitive function was evaluated by a trained geriatrician using the Mini Mental State Examination (MMSE) [32]. The medical history assessment included a record of conditions such as depression, anxiety, personality disorders, bipolar disorder, schizophrenia, obstructive sleep apnea, chronic obstructive pulmonary disease, heart failure and urinary incontinence. The ISI is a validated self-assessment scale that includes seven items, each rated on a 5-point scale (from 0 to 4), to measure insomnia severity. ISI scores categorise insomnia severity into the following levels: no insomnia (ISI: 0–7), mild insomnia (ISI: 8–14), moderate insomnia (ISI: 15–21) and severe insomnia (ISI: 22–28) [33]. The PSQI is a validated self-report questionnaire designed to assess sleep quality and disturbances across seven domains, each scored on a 0 to 3 scale, with a total score ranging from 0 to 21. Higher PSQI scores indicate poorer sleep quality [34]. Participants completed a 7-day sleep diary upon waking every morning. In the sleep diary, participants recorded nap times, bedtime, lights-off time, wake-up time, out of bed time, time to fall asleep and the number and duration of nighttime awakenings. Subjective measures of sleep onset latency (SOL), wake time after sleep onset (WASO), number of awakenings (NA), total sleep time (TST) and sleep efficiency (SE) were calculated from the sleep diaries. Objective data on WASO, NA, TST and SE were extracted from the actigraphy record from Actigraph® GT9X. All participant assessments were conducted by the principal investigator and trained research assistants.

Safety assessment

Weight, blood pressure, ECG, glycaemia, triglycerides, cholesterol, ALT, sodium, white blood cells and neutrophils count were measured at baseline and at the end of participation to monitor for potential side effects. An adverse reaction questionnaire assessing symptoms was administered via phone call at the end of week 2, 3, 4 and 5 by a trained research assistant. Detail safety assessment are available in supplementary materials.

Efficacy and safety endpoints

The primary efficacy endpoint was the mean difference in the ISI score from baseline. A reduction of six or more points in the ISI score is considered a minimally important difference for a clinically meaningful improvement in insomnia [35]. Yang et al. [35] estimated that a minimally important difference (MID) for a response in ISI score is a reduction of 6 points. The MID was estimated using 10 items from the Short Form Health Survey (SF-36), 3 from the work Limitation Questionnaire (WLQ) and the Fatigue Severity Scale (FSS). The vitality, role limitations due to physical health, and physical and social functioning items of the SF-36 are relevant to the older population. However, the WLQ may not accurately assess functioning in retired patients with insomnia. The secondary efficacy endpoints included the mean ISI score, the mean PSQI score at the end of the treatment period, as well as the mean subjective SOL, WASO, NA, TST, SE and the objective WASO, NA, TST and SE during week 5. The proportions of responders and remitters in each treatment group were also measured as secondary outcomes. Participants were classified as responders if they had a reduction of 6 or more points in their ISI score, and as in remission if their absolute ISI score was below 8 [35, 36]. Safety endpoints included any adverse events reported during the trial and all adverse events leading to the premature discontinuation of the study treatment. Adverse events were classified as severe if they resulted in death or hospitalisation.

Statistical analyses

The primary and secondary endpoints were analysed in the modified intent-to-treat population, defined as all participants who were randomly assigned that took at least one dose of the study medication. An analysis of covariance (ANCOVA) was used for primary endpoints, with the change in ISI score from baseline to day 28 as the dependent variable, baseline treatment group and baseline ISI score as an independent variable. Results are presented as estimates with adjusted 95% confidence intervals (95% CI), controlling for a two-sided 5% significance level. The exploratory secondary endpoints analysed means using a two-sided Student’s t-test, and proportions with a chi-squared test. Means were reported along with their standard deviations (SD). Missing data for endpoints were assumed to be similar to that of participants in the same treatment group. Missing data accounted for 15.8% of primary and secondary endpoints across all treatment groups. Sensitivity analyses using the baseline data carry-forward (BDCF) assumption were conducted to assess the robustness of results by carrying forward baseline data for participants lost to follow-up. All statistical analyses were performed using SPSS version 28 IBM software.

Sample size calculations were based on a two-sample F-test for ANCOVA. We estimated that a total sample size of at least 54 participants would be needed to provide 80% power to detect a 2.4-point difference in ISI scores between treatment groups (SD = 4; Effect size = 0.6) [37]. To account for discontinuation and loss to follow-up, we increased the sample size by 10%, resulting in a total of 60 participants.

Results

Study population

Overall, 231 participants were screened for inclusion between 1 October 2022 and 15 June 2024, of whom 60 (community: 45 and outpatient clinics: 15) were randomly assigned to mirtazapine or placebo (see Figure 2). Three participants withdrew their consent during the first week, prior to starting the research treatment, due to concerns about potential adverse events. The most common reasons for not meeting the inclusion criteria were current hypnotic use (n = 79; 54.1%), glaucoma (n = 15; 10.3%), and severe or untreated sleep apnea (n = 14; 9.6%). The mean age of participants was 72.3 ± 5.3 years, with the majority being female (n = 36; 60%). Participant characteristics are presented in Table 1. The prevalence of comorbidities was 88.3% (n = 53), with a mean of 3.4 ± 2.7 comorbidities and 56.7% (n = 34) had three or more comorbidities. Based on MADRS scores, 50% (n = 30) of participants had no symptoms of depression, 46.7% (n = 28) had mild symptoms and 3.3% (n = 2) had moderate symptoms. The average MADRS score was 8.5 ± 7.0 in the mirtazapine group compared to 6.5 ± 4.5 in the placebo group (P = 0.18). During the study, 71.7% (n = 43) of participants were taking medication, with an average of 3.8 ± 3.6 medications, and 18.3% (n = 11) were on five or more medications. No participant has received cognitive-behavioural therapy before entering the clinical trial.

Primary and secondary efficacy endpoints

Table 2 presents the results comparing the mirtazapine and placebo groups. The mean change in ISI score was −6.5 [95%CI: −8.3; −4.8] in the mirtazapine group and − 2.9 [95%CI: −4.4; −1.4] in the placebo group (P = 0.003, η_p² = 0.18). The proportion of responders was 50% (10 out of 20) in the mirtazapine group and 21.4% (6 out of 28) in the placebo group (P = 0.038). The proportion of participants in remission was 40% (8 out of 20) and 3.6% (1 out of 28) in mirtazapine and the placebo group respectively. In the sensitivity analysis using BDCF, the change of ISI score was −5.0 [95%CI: −4.2; −1.1] in the mirtazapine group and − 2.6 [95%CI: −4.2; −1.1] in the placebo group (P = 0.045). The mean PSQI score on day 28 (7.5 ± 2.8) was significantly lower compared to baseline (P < 0.001).

Safety endpoint

Adverse events affected 78.9% (n = 45) of the study population. The safety profiles are presented in Table 3. No participant experienced severe adverse events, and no deaths were reported during the study. Adverse events led to the discontinuation of 6 participants in the mirtazapine group (daytime drowsiness [n = 2; 33.3%], confusion [n = 2; 33.3%], hot flush [n = 1; 16.7%], blurred vision [n = 1; 16.7%]) and 1 participant in the placebo group (respiratory symptoms). No withdrawal syndromes were reported during the week following mirtazapine discontinuation. The mean weight gain in the mirtazapine group was 0.79 ± 1.46 kg, compared to 0.14 ± 1.60 kg in the placebo group (P = 0.22). No falls were reported. The average change in the corrected QT interval was 0.4 ± 15.2 ms in the mirtazapine group and 0.5 ± 20.3 ms in the placebo group. All QT readings at the study's end were below 485 ms.

Discussion

This study provides evidence of the efficacy of mirtazapine in reducing insomnia symptoms in patients with chronic insomnia. The improvement included a reduction in subjective WASO time, ISI score and an increase in TST and SE. In our study, 50% of participants achieved this minimally important difference, and 40% achieved remission after 28 days of treatment with mirtazapine 7.5 mg. The proportion of responders (50%) is comparable to that observed in a study on esmirtazapine in non-elderly adults (41%) [38].

The mean change of −6.5 points on the ISI score with mirtazapine is similar to the changes observed with orexin receptor antagonists, which ranged from −5.3 to −7.9 points [39–41], and with esmirtazapine (−6.8 points) [42]. These results cannot be directly compared to previous mirtazapine trials, as they did not include chronic insomnia patients or measure ISI score changes [22, 23]. However, comparing our findings to the ongoing DREAMING study could provide additional insights [43].

There were no statistically significant differences in the objective measurements of sleep parameters. However, the same trend in the reduction of WASO and improvement of TST and SE was observed. The lack of statistical significance is likely due to missing data, as participants frequently forgot to wear the device, leading to incomplete measurements.

Despite the absence of severe adverse events, a larger proportion of participants in the mirtazapine group withdrew from the study compared to the control group. This trend was also observed in the clinical trial with esmirtazapine [38]. The high incidence of adverse events in both the mirtazapine and placebo groups may be due to comorbidities and polypharmacy, which increased the risk of adverse events, whether or not they are related to mirtazapine [44]. Consequently, we assessed adverse events on a weekly basis, whereas other clinical trials typically assessed them every 1–2 months [39–41]. The incidence of daytime drowsiness and dizziness was high in the mirtazapine group, consistent with findings from a previous study on mirtazapine [45]. There was no significant weight gain associated with mirtazapine compared to placebo; however, this should be further investigated in longer clinical trials. Surprisingly, the incidence of dry mouth, increased appetite and weight gain did not differ from the placebo, despite these side effects being commonly associated with mirtazapine [45]. This may be due to the low dose and the short treatment duration in this study. A significant number of participants experienced flu-like symptoms whilst taking mirtazapine. Although these symptoms are not life-threatening, daytime drowsiness and flu-like symptoms may strongly limit the acceptability of mirtazapine for insomnia, warranting further exploration in future clinical trials.

Insomnia is associated to depression [46], and the DSM-V includes it as a criterion for major depressive disorder [47]. Although participants were excluded for a major depressive disorder, assessing subclinical depressive symptoms was important for characterising the study population. The presence of depressive symptoms in half of the population suggests mirtazapine may be effective for both patients with and without depressive symptoms. This hypothesis could be explored in future clinical trials.

Given that mirtazapine is an antidepressant, it is also relevant to examine depressive symptoms to explore whether its hypnotic effects are related to its antidepressant properties. Due to the presence of depressive symptoms in some participants and the rapid onset of action of mirtazapine [48] its efficacy in reducing insomnia symptoms could potentially be partially mediated by its antidepressant effect. Depressive symptoms could have been assessed at the end of the treatment period for efficacy analysis as the primary objective was to evaluate mirtazapine's effect on insomnia. The small sample size limited the inclusion of an additional covariate, but exploring mirtazapine's potential antidepressant effects on its hypnotic properties is relevant for future trials.

The double-blind, placebo-controlled design strengthens the validity of our results compared to previous studies on the efficacy of mirtazapine for chronic insomnia [22, 28]. A key strength of this study is the use of a validated insomnia severity scale, which provides a more comprehensive assessment of mirtazapine's efficacy beyond sleep data alone. However, several factors limit the interpretation of our findings. The data on the efficacy and safety of mirtazapine covers only a 4-week period, which may limit the study’s generalisability, especially given that older patients often use hypnotics for longer durations. The use of a low dose of mirtazapine for treating insomnia is also a limitation. Nonetheless, our study remains relevant, as its follow-up time is longer than most published clinical trials [49]. However, mirtazapine’s efficacy and safety should still be assessed over a period longer than 28 days. The proportionally large loss to follow-up in our study is a substantial limitation. A total of 37% of participants were lost to follow-up in the mirtazapine group, compared to 13% in the placebo group. Although the pessimistic sensitivity analysis (BDCF) still demonstrated a statistically significant reduction of insomnia symptoms, this imputation method does not totally eliminate the study’s power limitations. This underscores the relevance and need for a larger clinical trial to confirm the hypnotic effect and safety of mirtazapine. A large population of adults aged 65 and older may benefit from mirtazapine, as this study included a diverse range of participants with and without comorbidities and concomitant medications. However, CBTi remains the first line treatment for insomnia, and efficacy and safety data might differ in a younger population.

Conclusion

In conclusion, mirtazapine 7.5 mg may be considered a therapeutic option for the treatment of chronic insomnia in older adults. It demonstrated a clinically significant effect on self-reported insomnia symptoms without causing severe adverse events. However, adherence to mirtazapine may be challenging, as indicated by a higher discontinuation rate due to adverse events. Given the loss to follow-up, further studies are needed to validate the short-term efficacy and safety of this treatment, as well as to assess its long-term effects.

Acknowledgements:

We would like to express our sincere gratitude to Lynda Rabia who has contributed to the recruitment and data collection of study participants.

Declaration of Conflicts of Interest:

TDV has received consultant fees from Eisai and Idorsia, and research grants from Paladin Labs and Jazz Pharmaceuticals. DP has received consultant fees from Eisai.

Declaration of Sources of Funding:

This work was supported by the Quebec Research Network on Ageing.

References