According to the World Health Organization more than 300 million people in the world suffer from depression, which is the single largest factor contributing to global disability.
(Smith & De Torres, 2014) During the last three decades, the number of cases of depression worldwide increased by almost 50%, with growing age-standardized incidence rates in regions of any sociodemographic index. (Liu et al., 2019) Depression is and remains a major public health problem.
Antidepressants are a mainstay of treatment for major depression but are associated with clinically relevant risks of severe adverse events during both use and withdrawal. (Henssler et al., 2019) To increase the efficacy of treatment, inpatients could potentially benefit from comprehensive, non-pharmacological elements of therapy. (Qaseem et al., 2016, Zaprutko et al., 2020) Among these, relaxation techniques are commonly used to treat depression and have proven effective both solely and adjunctively to standard care. (Af et al., 2010, Jia et al., 2020)
Along with the rise of mindfulness-based interventions in mental health care, meditation has received growing attention in research and clinical practice. Meditation, although often used synonymously with “mindfulness meditation” is an umbrella term for a large variety of techniques and definitions. (Matko & Sedlmeier, 2019)
Meditation usually involves sustained attention on an object, which can be external, or a physical or mental process. Meditative techniques exceed the positive psychological effects of mere relaxation techniques in healthy individuals (Sedlmeier et al., 2012, Sedlmeier et al., 2018) and recent meta-syntheses of randomized controlled trials unanimously concede that meditation has substantial beneficial effects on depressive symptoms in clinical populations. (Goldberg et al., 2020, Rose et al., 2020) However, the results so far were quite heterogeneous: depending on type of meditation, population, and outcomes the effects of meditation were not reliable or superiority could not be demonstrated compared to active control groups. (Goldberg et al., 2020, Goyal et al., 2014)
Research on mantra meditation in clinical, but also in other, contexts is still in its infancy. Compared to mindfulness meditation, there are still relatively few studies and these are predominantly characterized by lack of methodological rigor. (Goyal et al., 2014, Ospina et al., 2007) Equally relevant, however, is the fact that there is still no overarching theory on the mechanisms of action of mantras in a scientific context. Such a theory should be informed by the traditional and spiritual contexts of mantra meditation—so far, mantra meditation has mainly been examined in a secularized form by depriving meditation techniques of their traditional background (e.g., relaxation response, (Benson et al., 1974) clinically standardized meditation (Carrington et al., 1980)), or by examining singular techniques in their own context (e.g., transcendental meditation, (Ospina et al., 2007) mantra repetition program (Bormann et al., 2014)).
As an example, a recent mantra meditation program in a clinical setting that is based on the traditional background of classical yoga is Meditation Based Lifestyle Modification (MBLM). (Bringmann et al., 2020, Bringmann et al., 2021) Classical yoga is oriented toward spiritual liberation through meditation. It provides principles and practice of an ethical lifestyle, body-oriented yoga practices, and meditation in order to achieve expanding stillness of the mind and, eventually, self-realization.
Even more important than in yoga, mantra meditation is a key feature of Tantrism. Tantrism is an umbrella term for ancient esoteric traditions of Hinduism and Buddhism, in which highly differentiated mantras (e.g., specific sound formulas) and yantras (e.g., specific geometrical patterns) are used to reproducibly achieve various worldly and transcendent goals. (Feuerstein, 1998, Ramachandra Rao, 1979) In India, a “spiritual science” has developed concerning the employment and utility of mantras, (Ramachandra Rao, 1979) including initiation by a spiritual teacher and specific rules and regulations for practice. (Sivananda and Yoga, 2010) A mantra is described as a “mystical energy encased in a sound structure,” with deep faith in the mantra being paramount to unleashing its effects on the body, mind, and soul. (Devananda, 1995, Sivananda and Yoga, 2010) These esoteric and mystical concepts were largely ignored in Western research (e.g., in clinical sciences), or mantras were disregarded as primitive magico-religious practices (e.g., spells or prayers). However, Western scholars predominantly arrived at these interpretations from Christian theology, which led to misleading ideas that have strongly influenced discussions of Indian mantras over the last few decades. (Fárek & Horák, 2021) In contrast, the knowledge on mantra and yantra (Feuerstein, 2012) is more than five millennia old. Traditional practices and their metaphysical assumptions thus deserve serious and open-minded consideration in future research.
While most meta-analyses report on mindfulness meditation, the evidence for the effects of mantra meditation—a common meditation technique among adults in the United States (Burke et al., 2017)—is still scarce. An earlier meta-analysis (Goyal et al., 2014) in 2014 reported either insufficient or low evidence for a positive effect of mantra meditation on anxiety, depression, stress, quality of life, and sleep in clinical populations. However, as only a few mantra meditation programs met the inclusion criteria for this analysis, the authors’ ability to draw conclusions about the effects was significantly limited. Furthermore, the authors reported a likely “floor” effect in selected samples of the general population, if symptom levels of the outcomes were low before the intervention. (Goyal et al., 2014) Another meta-analysis in 2014 on trait anxiety (which is strongly correlated with depressive symptoms), (Knowles & Olatunji, 2020) however, reported transcendental meditation (a mantra-based meditation technique) to be more effective than most active control treatments. (Orme-Johnson & Barnes, 2014) In a recent systematic review of mantra meditation for mental health in the general population, (Lynch et al., 2018) eight of 11 studies reported significant improvements on outcomes of depression (four randomized controlled trials and four other designs).
Overall, the existing reviews illustrate the lack of evidence on the efficacy of mantra meditation in mental health care due to the limited available studies, both in number and methodological rigor. This study aims to add evidence for the effects of mantra meditation in inpatients with major depression compared to progressive muscle relaxation (PMR) by assessing measures of clinician-rated and self-reported depression over six months after discharge from hospital. A Cochrane review has shown variations of PMR to be more effective at reducing self-rated depressive symptoms than no or minimal treatment; (Bringmann et al., 2020) it may add additional efficacy to antidepressant treatment. (Bringmann et al., 2021) Also, we know from meta-analyses of effects of meditation in healthy individuals, that relaxation techniques like PMR are regularly chosen as an active control group. (Sedlmeier et al., 2012, Sedlmeier et al., 2018) Thus, PMR was considered a valid active control group in this study.
This two-arm, single-site, rater-blinded, randomized controlled trial (ClinicalTrials.gov Identifier: NCT03004430) was conducted at the Clinic for Psychiatry, Psychosomatics, and Psychotherapy, Zschadrass, Germany from January 2017 through June 2020. The study was approved by the ethics review board of the Technische Universität Dresden (EK 498,122,016). All participants provided written informed consent after receiving a full explanation regarding the study protocol and before randomized assignment to treatment arms. The study employed open allocation (participants were informed of assignment after randomization) but included blinded assessment of the primary outcome measure before treatment (T0), at hospital discharge (T1), at 3-month follow-up (T2), and at 6-month follow-up (T3).
Inclusion criteria were inpatient treatment, a diagnosis of a current episode of major depression as defined by DSM-IV criteria (DSM-IV codes: 296.22, 296.23, 296.32, 296.33. Corresponding ICD-10 codes: F32.1, F32.2, F33.1, F33.2), age ≥18 years, a score of 20 or greater in the Beck Depression Inventory (BDI-II), (Beck et al., 1996) and the ability to sit on a chair for at least 20 min.
Patients were excluded for any one of the following reasons:
1. Any psychiatric comorbidity except abuse of or dependence on nicotine, agoraphobia with or without panic disorder, generalized anxiety disorder, posttraumatic stress disorder, specific phobia, social phobia, hypochondriasis, pain disorder, somatization disorder, undifferentiated somatoform disorder, sexual and gender identity disorders or eating disorders
2. Psychotic symptoms not reconcilable with unipolar depression
3. acute suicidality
4. diagnoses affecting cortisol levels such as type I diabetes mellitus, cancer, asthma, chronic hepatitis, chronic fatigue syndrome, or regular use of medication with an immunomodulatory effect (e.g., cytotoxic chemotherapy, corticosteroids, interferons)
5. The current practice of other forms of mantra repetition such as the rosary, chanting, or transcendental meditation; 6. current participation in another clinical trial.
2.3. Primary and secondary outcome measures
The primary outcome measure was the change in the score on the Montgomery and Åsberg Depression Rating Scale (MADRS). (Schmidtke & Moises, 2007) This clinician-rated, 10-item scale is designed to be sensitive to the change in effects of depressive symptoms during drug therapy. The internal consistency of MADRS is considered very high and the correlation of MADRS has been shown to be generally high or very high with the Hamilton Rating Scale for Depression (HAM-D). (Hamilton, 1960) A score greater than 30 or 35 on MADRS indicates severe depression, while a score of 10 or below indicates remission. To maintain high levels of interrater reliability in this study, trained psychologists used a structured interview guide (SIGMA) (Williams et al., 2008) that has been developed for MADRS.
The secondary outcome measure was the change in the score on the Beck Depression Inventory (BDI-II). (Beck et al., 1996) This is a 21-question multiple-choice self-report inventory, one of the most widely used psychometric tests for measuring the severity of depression. The BDI-II is composed of items relating to symptoms of depression such as hopelessness and irritability, cognitions such as guilt or feelings of being punished, as well as physical symptoms such as fatigue, weight loss, and lack of interest in sex. Cut-off scores: values below 14 can be considered as subclinical depression; 14–19 – mild depression; 20–28 – moderate depression; and 29–63 – severe depression. The present study focuses on depression. Other variables assessed in this research project, including parameters of heart rate variability and measures of spirituality, will be discussed in separate publications.
Patients with suspected depression were screened consecutively at hospital admission by administration of the BDI-II questionnaire. In a personal interview within 72 h of admission, a senior psychiatrist assessed those with a BDI-II score ≥ 20.
Consenting participants were examined for eligibility, utilizing the Structured Clinical Interview for DSM-IV (SCID) (First et al., 1996) to verify DSM-IV criteria of major depression and exclusionary conditions. Patients meeting the study criteria took part in the MADRS interview and completed questionnaires for clinical measures, demographic data, and medication. They were then enrolled prior to randomization, which was implemented by a software-based minimization algorithm, (O’Callaghan, 2014) stratified for gender, age, diagnosis (first or recurring depressive episode), severity of depression (baseline BDI-II score), and religious affiliation (none vs. any) with 20% randomization. Within these 20%, subjects were randomized 1:1 to either treatment.
Study physicians or psychologists who were trained by standardized procedures to establish interrater reliability gave the MADRS ratings. To assess interrater reliability, 15% of total number of interviews (n = 65) were rated by two raters simultaneously. The average intraclass correlation coefficient was 0.99 with a 95% confidence interval from 0.98 to 1.0 (F(64, 65) = 179, p<.001), which reflects a high degree of reliability.
2.5. Treatment conditions
The treatment group received training in mantra meditation (MAM, see below), and the control group received training in progressive muscle relaxation (PMR, see below). Both groups did their training in an introductory class, followed by 30-minute group sessions held twice a week. Participants in both groups used standardized manuals, instructor guides, information brochures, and online content with guided audio sessions for personal use. The MAM and PMR therapy facilitators each facilitated only one treatment type. Nursing staff with long-term experience in PMR provided this therapy, and experienced meditation practitioners led the mantra meditation. Participants were encouraged to practice on their own around 20 min per day and to keep a record of their practice in a diary provided in the study brochure. Both treatments were delivered adjunctively to treatment as usual (TAU), which included an individual program of pharmacotherapy, psychotherapy, ergotherapy, and movement therapy as part of standard inpatient care. After discharge, participants could choose to receive weekly, group-specific emails as a reminder and incentive for continued practice. Participants were asked to refrain from practising other methods of relaxation or meditation during the study.
2.6. Mantra meditation
In mantra meditation, a spiritually related sound, word, or phrase is silently repeated with one-pointed attention toward the mantra. Participants could choose from mantras from a variety of spiritual traditions during the introductory class.
The facilitators advised participants as follows: They were to choose a pleasant-sounding mantra that was compatible with their personal beliefs and easy to remember (e.g., “Ave Maria”, “Om mani padme hum”, “Om nama shivaya”). During meditation, participants could sit in any, preferably upright, position. The mantra should be recited inwardly with focused attention on the sound rather than its meaning. However, appreciation of the mantra’s meaning and belief in it may act as a catalyst for meditative practice. (Wachholtz, 2008) Synchronization with the breath was not part of the technique, but was allowed to occur spontaneously or at will. Arising thoughts should not be judged or followed. With growing meditation practice, concentration on the mantra may become increasingly receptive, sustained, and effortless, allowing for deeper states of experience. (Telles et al., 2016) As an optional technique, participants could choose and visualize a geometrical pattern (yantra) along with the recitation of their mantra, combining auditory and visual concentration (e.g. a symbol of the associated spiritual tradition or simple geometric patterns such as a triangle with a central point in the middle). Also, facilitators encouraged participants to use mantra recitation as a tool for self-regulation in situations of distress in daily life. A more detailed description of the meditation technique, including a list of mantras, yantras, and different systems for classifying meditation is provided in the supplementary material.
2.7. Progressive muscle relaxation
PMR was introduced in 1934 by Jacobson to counter feelings of anxiety and distress mediated by sustained muscle tension. (Jacobson, 1938) The technique involves first tensing and then releasing different muscle groups of the body, with attention being directed toward the differences perceived during tension and relaxation. PMR is considered as an adjunctive therapy for depression and can provide patients with self-maintenance coping skills to reduce depressive symptoms. (Jorm et al., 2008)
2.8. Data analysis
The primary objective was to test differences between groups (active control: TAU + PMR; experimental: TAU + MAM) in total scores of MADRS after inpatient treatment, after 3 months, and after 6 months (time x group interaction). For sample size calculation to detect this interaction parameter, there were no sufficient previous studies we could refer to. Although potentially too limited to be fully reliable, we estimated expected mean values in MADRS scores by taking into account three longitudinal studies on the course of depression over 6 months during psychopharmacological treatment. (Allard et al., 2004, Montgomery et al., 1993, Trick et al., 2004) We calculated the sample size using GLIMMPSE, (Wang et al., 2013) resulting in total sample size of N = 108, e.g., n = 54 within each group with the following assumptions: a α-level of α = .05, a power of (1-β) = .80 and a correlation among repeated measures of r = .5. The resulting estimated difference of means at 3-month and 6-month follow-up was higher or within the range of minimal clinically important difference (MCID) of MADRS, which has been reported to range from 1.6 to 1.9. (Duru & Fantino, 2008) We adjusted the sample size for a 20% attrition rate due to loss to follow-up or dropout. With these assumptions, we calculated a sample size of 130. We did not conduct interim analyses, but recruitment for the study ended in February 2020, due to shortage of care related to the COVID-19 pandemic, prior to reaching the predetermined recruitment goal (final N = 123 participants).
To summarize the baseline sociodemographic variables, we used means for quantitative variables and proportions for categorical variables. To check for balance between the study groups, we compared demographic data and other important baseline characteristics with independent-samples t tests for continuous variables and χ2 tests for categorical variables.
In the primary analysis, we compared change in depression severity using MADRS and BDI-II scores at the posttreatment and follow-up assessments. First, we used a random-effects linear mixed model (LMM) to examine the overall intervention impact (i.e., overall effect of visits and study treatment group two-way interactions at all post-baseline visits) and then to separately evaluate changes in MADRS and BDI-II scores between treatment groups across four time points from admission to 6 months after discharge from hospital. The model included participants as a random effect, treatment and time as fixed within-subject factors, interaction between treatment and time, and a random intercept. As recommended for longitudinal analysis of RCTs, (Carli et al., 2020) we included baseline scores as covariates with fixed within-subject and between-subject effects, respectively. We included time as a categorical covariate with baseline as a reference category. In an exploratory analysis, we grouped items of primary and secondary measures across both scales according to three factors presented in an exploratory factor analysis of MADRS, BDI-II, and HAMD-D in the larger-scale GENDEP study (observed mood, cognitive, and neurovegetative). (Uher et al., 2008) We used independent-samples tests to calculate and compare changes from baseline to 6-month follow-up .
All participants, irrespective of treatment compliance, were included in the analysis according to their allocated treatment group at randomization. We computed effect sizes as Cohen’s d and an effect size parameter (Westfall et al., 2014) for the LMM. In statistical tests, we used a two-sided benchmark of p-values less than 0.05 as the general error rate for significance. We ran all analyses in R 3.6.
3.1. Descriptive analysis
Table 1. Demographic and clinical characteristics of study participants by group.
|Age||123||0||Mean (SD)||45.0 (14.0)||45.9 (12.5)||0.721b|
|Gender||123||0||Female||43 (71.7)||40 (63.5)||0.438b|
|Male||17 (28.3)||23 (36.5)|
|Marital status||121||2||Unmarried||18 (30.0)||20 (31.7)||0.939c|
|Divorced||10 (16.7)||8 (12.7)|
|Married||28 (46.7)||31 (49.2)|
|Widowed||3 (5.0)||3 (4.8)|
|Current occupation||121||2||Self-employed||1 (1.7)||2 (3.2)||0.423c|
|Employed||31 (51.7)||43 (68.3)|
|Unemployed||12 (20.0)||8 (12.7)|
|Pension (disability)||10 (16.7)||6 (9.5)|
|Pension (old-age)||4 (6.7)||4 (6.3)|
|Country of Origin||123||0||Germany||59 (98.3)||61 (96.8)||1.000c|
|Other||1 (1.7)||2 (3.2)|
|Denomination||122||1||Catholic||0 (0.0)||2 (3.2)||0.457c|
|Protestant||7 (11.7)||10 (15.9)|
|None||49 (81.7)||49 (77.8)|
|Other||3 (5.0)||2 (3.2)|
|Cigarettes per day||78||45||Mean (SD)||5.8 (8.7)||3.5 (6.3)||0.166b|
|Alcoholic drinks per week||72||51||Mean (SD)||1.3 (3.5)||1.7 (3.0)||0.564b|
|Depression type||123||0||First Episode||28 (46.7)||32 (50.8)||0.782c|
|Recurrent episodes||32 (53.3)||31 (49.2)|
|ICD-10 F diagnoses||123||0||1||37 (61.7)||43 (68.3)||0.517c|
|2||18 (30.0)||13 (20.6)|
|3||5 (8.3)||6 (9.5)|
|4||0 (0.0)||1 (1.6)|
|ACE||106||17||< 4||32 (61.5)||42 (77.8)||0.108c|
|≥ 4||20 (38.5)||12 (22.2)|
|Inpatient treatment (days)||123||0||Mean (SD)||40.0 (17.2)||39.5 (14.7)||0.847b|
|Max. care after discharge||94||29||Pharmacotherapy||8 (13.3)||6 (9.5)||0.413c|
|Psychotherapy||4 (6.7)||4 (6.3)|
|Integrated outpatient care||10 (16.7)||11 (17.5)|
|Day clinic||19 (31.7)||29 (46.0)|
|Inpatient treatment||0 (0.0)||3 (4.8)|
P-value of the comparison between the two groups.b
χ2 for independence test.
The length of inpatient treatment was 39.5 days for the PMR group (SD: 14.7), and 40.0 days in the MAM group (SD: 17.2). Participants reported practicing an average of 14.85 min per day during the whole study period. The mean amount of practice did not differ between the groups (ΔM=1.69, 95% CI (−0.45,3.83), t(247.30)=1.56, p=0.120). Most patients opted for the weekly email reminders after discharge (115 of 123 participants, 93%) with no significant difference in group.
Attrition after randomization did not differ significantly between the treatment arms, although participants in the MAM group had numerically higher rates of attrition than those in the PMR group at discharge (8% vs. 3%; χ2 = 0.71, p = .398), at the 3-month follow-up (35% vs. 21%; χ2 = 2.33, p = .114), and at the 6-month follow-up assessment (42% vs. 27%; χ2 = 2.33, p = .127). (see Fig. 1)
There were no significant differences in baseline characteristics between study completers and dropouts within each group and between groups (Fig. 2).
3.2. Primary and secondary outcome measures
The baseline-adjusted analysis of the primary outcome revealed significant main effects for time (F(3,311.55)=212.65, p<0.001) and group (F(1,129.18)=10.02, p=0.002) and. more importantly, a significant interaction between time and group (F(3,311.61)=4.75, p=0.003). This interaction was qualified by a mean decline in MADRS score from baseline to 6-month follow-up (baseline unadjusted) of 20.6 points in participants allocated to TAU + MAM treatment, as compared to 15.4 points in the TAU + PMR group. Baseline-adjusted treatment contrasts revealed no between-group difference in mean change from baseline to hospital discharge (p = .126, 95% CI: −5.61 – 0.69), but significant between-group differences at both follow-ups (3-month follow-up: p = .006, 95% CI: −8.32 – −1.42; 6-month follow-up: p = .001, 95% CI: −9.80 – −2.69). The baseline-adjusted analysis of the secondary outcome led to significant main effects for time (F(3,274.51)=137.15, p<0.001), but not for group (F(1,125.55)=0.80, p=0.372) nor for interaction between time and group (F(3,274.53)=0.33, p=0.801). The mean decline of the change in BDI-II scores from baseline to 6-month follow-up (baseline unadjusted) was 19.6 points in participants allocated to TAU + MAM treatment as compared to 16.6 points in the TAU + PMR group. Baseline-adjusted treatment contrasts revealed no significant between-group difference in changes from baseline to any later visit (hospital discharge: p = .446, 95% CI: −5.62 – 2.47; 3-month follow-up: p = .636, 95% CI: −5.55 – 3.39; 6-month follow-up: p = .371, 95% CI: −6.80 – 2.54). Still, the intervention impact showed in the same direction as for the primary outcome and was estimated to decrease to −2.1 at 6-month follow-up (Table 2 and Fig. 2).
Table 2. Primary and Secondary Outcomes.
|Outcome||Time||Mean (SD)||dc||Impact||95% CI||p||d|
|MADRS||Baseline||28.4 (6.2)||30.2 (5.2)|
|Discharge||12.2 (8.5)||11.4 (7.3)||.11||−2.46||−5.61 – 0.69||.126||.35|
|3-month||13.5 (10.1)||10.7 (7.4)||.32||−4.87||−8.32 – −1.42||.006||.62|
|6-month||14.0 (11.5)||9.6 (7.8)||.45||−6.24||−9.80 – −2.69||.001||.79|
|BDI-II||Baseline||31.7 (8.9)||32.9 (8.0)|
|Discharge||14.9 (11.7)||13.8 (10.0)||.10||−1.57||−5.62 – 2.47||.446||.19|
|3-month||15.3 (12.4)||14.5 (11.2)||.07||−1.08||−5.55 – 3.39||.636||.13|
|6-month||15.1 (12.0)||13.3 (12.3)||.15||−2.13||−6.80 – 2.54||.371||.26|
Note. BDI-II: Beck Depression Inventory; MADRS: Montgomery-Åsberg Depression Rating Scale score; MAM: Mantra Meditation; PMR: Progressive Muscle Relaxation.
Overall: time by group interaction; SD: Standard Deviation; dc: Cohen’s d; Impact: model estimate; CI: Confidence Interval; d: effect size parameter.63.
3.3. Exploratory analysis
To further understand the differences between the primary and secondary outcomes, we evaluated pre-post differences (e.g., 6-month follow-up vs. baseline) of each item of both questionnaires (MADRS and BDI-II) by treatment group and with respect to the following factors: 1) observed mood, 2) cognitive, and 3) neurovegetative (Fig. 3). Compared to participants practicing PMR, mantra meditators showed more improvements in observed mood items than in cognitive items (ΔM=0.09, 95% CI (0.04,0.13), t(12.78)=3.96, p=0.002) and, tended to show more improvements in neurovegetative items (ΔM=0.08, 95% CI (−0.01,0.16), t(7.53)=1.98, p=0.086). Improvements between cognitive and neurovegetative items did not differ significantly (ΔM=0.01, 95% CI (−0.08,0.10), t(6.45)=0.27, p=0.795).
In this randomized controlled trial, 123 inpatients with major depression received relaxation- or meditation-based interventions adjunctively to treatment as usual (individualized pharmacotherapy, psychotherapy, ergotherapy, and physical activity). Mantra meditation was associated with greater reductions in the clinician-rated primary outcome measure (the MADRS) at the 3-month and 6-month follow-up assessments than PMR, with a medium effect size at both time points. Mantra meditation was not associated with greater reductions in self-reported BDI-II scores than PMR at any assessment. An exploratory analysis revealed that the superiority of the mantra meditation group regarding the primary outcome was mainly determined by mood and neurovegetative, rather than cognitive symptoms.
This is the first randomized trial with an active comparison treatment to examine the efficacy of adjunctive mantra meditation in inpatients with major depression. As stated in the introduction, existing evidence on the efficacy of mantra meditation in depression is non-conclusive. In populations without clinically diagnosed depression, some randomized trials reported no effect of mantra meditation compared to an active control group; (Alexander et al., 1989, Chhatre et al., 2013, Jayadevappa et al., 2007, Schneider et al., 2012) others report small to moderate effect sizes. (Bormann et al., 2006, Bormann et al., 2018, Wolf & Abell, 2003) Effect sizes reported from randomized trials conducted with a waiting-list control tend to be higher, but also range from small to medium. (Elder et al., 2014, Leach et al., 2015, Nidich, 2016, Nidich et al., 2009) Our findings add to the literature showing that mantra meditation can yield substantial improvements in depressive symptoms. Effect sizes reported in this study were in the upper range of prior results, especially when considering the adjunctive character of the interventions. Nevertheless, there was no difference in depression levels at discharge, which shows that at this point, the TAU outweighs the additional effects of meditation. Follow-up data revealed that the effect of mantra meditation is more stable (or even grows) in the longer term, which is in line with previous findings, that effects of meditation increase with periods of intervention exceeding 6 weeks. (Sedlmeier et al., 2018) Since both groups received the same support for continued practice and the amount of practice did not differ significantly between groups, these effects can be attributed to mantra meditation itself. Effects were clinically important: assuming a conservative MCID of 1.9 for the MADRS, (Duru & Fantino, 2008) improvements in primary outcome measure are higher than those estimated by the baseline-adjusted model by factors of 1.29 (discharge), 2.56 (3-month follow-up), and 3.28 (6-month follow-up). For the secondary outcome measure, mantra meditation did show numerical improvements compared to the control group, but these remained lower than the statistically significant and clinically important effects reported for BDI-II, (Button et al., 2015) except for the last follow-up (factors of 0.87 at discharge, 0.56 at 3-month follow-up, and 1.13 at 6-month follow-up). Our exploratory analysis revealed that mantra meditation during a study period of 6 months had higher impact on observed mood than on cognitive symptoms. This adds to the literature showing higher effects of concentrative meditation (versus mindfulness meditation) on negative emotions, anxiety, and neuroticism. (Sedlmeier et al., 2012, Sedlmeier et al., 2018) It also supports the understanding of meditation as disengagement of analytic or cognitive processes of the mind, which are often dysfunctional in depressive patients. In this interpretation, disengagement of dysfunctional cognitive patterns leads to better mood as an early effect of meditation, while the long-term goal of meditation would be to completely disidentify with these patterns. (Bryant, 2009) The analysis also demonstrated a tendency to lower effects of mantra meditation on neurovegetative symptoms than on observed mood. Although meditation can be understood as a neurophysiological bottom-up process regulating stress responses, (Gard et al., 2014) this intervention did not include important factors like breathing control and physical postures. Usually, only in long-term practitioners reach deep states of relaxation and absorption, underlining the visibility of early effects of mantra meditation in this study.
In our study design, we addressed frequent methodological issues in mind-body medicine research, including the absence of randomization or an active control group. Strengths of our study included clinician-blinded ratings of depression (the MADRS) and the complementary use of patient-reported BDI-II, clinically and theoretically important exploratory analyses, sequential screening of all admissions to hospital to avoid selection bias, stratification of our randomization by relevant patient characteristics, the inclusion of baseline depression in the linear mixed models, collection of data concerning interrater reliability and frequency of meditation/relaxation practice, and treatments of equal intensity using standardized guidelines. As a side note, the TAU + PMR group was likely to receive even more guided PMR sessions after discharge, if participants were treated in day clinics or other forms of integrated care. Mantra meditation, however, is usually not available in these settings, leading to a potential disadvantage for the TAU + MAM group.
Limitations of this study include smaller sample size and higher dropout rates than anticipated, a lack of session fidelity assessments, and the use of patient reports to investigate practice frequency over time. Since we did not implement a waiting-list control condition, it is theoretically possible that in this study, the control treatment might have been unintentionally delivered in a less effective way than reported in former studies. (Jorm et al., 2008) This is, however, unlikely, as PMR has been a standardized and well-established therapy for many years at the trial site and was provided by experienced facilitators. Attrition after randomization did not differ significantly between the treatment arms, although participants in the meditation group had numerically higher rates of attrition than those in the control group. However, there were no significant baseline differences among study completers and dropouts in each group and between groups. Also, the linear mixed models used all available data on study completers and dropouts alike to avoid biased results toward the mantra meditation group. Finally, as both interventions were delivered openly to the patients, the results may have been biased by the patient’s expectations.
For inpatients with major depression, adjunctive mantra meditation was found to greater and clinically relevant reduction in depressive symptom severity than progressive muscle relaxation. Mantra meditation mainly led to improvements in mood, followed by improvements in neurovegetative and then cognitive symptoms of depression. These promising results, the untapped potential of traditional aspects of mantra meditation, and the current scarcity of methodologically sound trials make further exploration of mantra-based meditation undoubtedly desirable.
Credit authorship contribution statement
Holger C. Bringmann: Conceptualization, Data curation, Writing – original draft, Investigation, Formal analysis, Funding acquisition. Aline Sulz: Data curation, Writing – review & editing. Philipp Ritter: Supervision, Writing – review & editing. Stefan Brunnhuber: Supervision. Michael Bauer: Supervision. René Mayer-Pelinski: Formal analysis, Supervision, Writing – review & editing.