Introduction

Mindfulness-based interventions (MBIs) integrate mindfulness principles and mindfulness meditation practice in established psychotherapies (eg, cognitive behavioral therapies), and other health interventions (eg, stress reduction, mindful eating, and mindful physical activity) [-]. Despite differences in theoretical perspective and the extent of formal mindfulness meditation practice, MBIs emphasize development of (1) attention regulation, including moment-to-moment awareness of internal and external experiences and (2) an open, nonjudgmental orientation toward these experiences [-]. MBIs have demonstrated effectiveness across multiple psychological outcomes (eg, anxiety, depression, pain, etc), populations (eg, children, youth, and adults), and delivery modes (eg, face-to-face and online), with Cohen d effect sizes ranging between 0.10 and 0.89 [].

The search for mechanisms of action (ie, active ingredients), in addition to evaluations of clinical effectiveness, is essential for a thorough assessment of MBIs. Mediation analyses, which are widely applied in evaluations of physical and mental health interventions [-], offer a useful framework for examining patterns of direct and indirect (ie, mediated) interrelations among predictors and outcomes. Accordingly, mediation analyses play a critical role in the development of empirically driven theories of how interventions deliver benefits.

Previous efforts in identifying mediators of MBIs indicate that various psychological constructs account for beneficial mental health impacts. Specific mediators, across MBIs following different theoretical perspectives, include: mindfulness [-], self-compassion [,,,,], decentering [,,], psychological flexibility [,], and acceptance [,,], in addition to reduced rumination [,,], worry [,,], experiential avoidance [,], cognitive defusion [], anxiety sensitivity [], positive or negative affect [], dysfunctional attitudes [], and cognitive-emotional reactivity [,,,].

Mindfulness, as a key mechanism of action in MBIs, is regarded in current research as both a disposition [,], and a capacity to regulate attention and emotion in relating to experiences with an open, accepting, and nonavoidant attitude [,]. This capacity, developed in mindfulness meditation and mindfulness skills training, has been differently operationalized in current literature [,]. For example, the Five Facet Mindfulness Questionnaire (FFMQ) [], a widely used measure of mindfulness, operationalizes mindfulness as a multidimensional construct, consisting of five conceptually distinct but inter-related facets. These include: the observing (OBS), describing (DES), acting with awareness (AAW), nonjudgment (NJU), and nonreactivity (NRE) facets. These facets show different patterns of association with indices of mood and sensitivities to change in response to treatments [-]. However, relatively few studies have examined mindfulness facets, as conceptualized in the FFMQ, as mediators of MBIs [-,]. Further examination of FFMQ-SF facets with respect to differences in sensitivity to change, norms in different populations, and mediating influence is warranted.

Using data from the Mindfulness Virtual Community (MVC) intervention, this study aims to evaluate the role of FFMQ facets as mediators of intervention effects on depression, anxiety, perceived stress, and quality of life outcomes. The MVC intervention was an 8-week web-based intervention based on mindfulness and cognitive-behavior therapy (M-CBT) principles that addressed psychological distress in university students [,]. This intervention comprised (1) a total of 16 psycho-educational and mindfulness meditation instruction videos (with associated printed text), (2) an anonymous peer discussion forum, and (3) an anonymous counselor-moderated video discussion session. Results from previous randomized controlled trials (RCTs) of the MVC intervention (2017‐2019) have been published [-], demonstrating effectiveness at improving variables reflecting depression [,], anxiety [,], mindfulness [,], quality of life [], and perceived stress [,], when compared to waitlist control (WLC) groups.

MethodsParticipants

Participants were undergraduate students from a large university in Toronto, Ontario, Canada, who participated in the MVC intervention during the 2017 and 2018 academic years. Details on study design, recruitment, procedures, and psychometric questionnaires have been previously published [-]. Given identical recruitment and program administration in the 2 study cohorts (2017‐2018), data from the 2017 and 2018 cohorts were aggregated for this analysis. Data from a third RCT, which was undertaken during a university-wide labor strike in 2019, were not included in this analysis.

Measures

Participants were administered several questionnaires, at baseline and at 8-week post intervention: (1) The Five Facet Mindfulness Questionnaire-Short Form (FFMQ-SF), a 24-item, 5-point Likert scale (item range: 1‐5, total score: 24‐120), which evaluated 5 mindfulness facets []. These included the OBS (4 items), DES (5 items), AAW (5 items), NJU (5 items), and NRE (5 items) facets. Depression was assessed by the Patient Health Questionnaire-9 (PHQ-9) [], a 4-point Likert scale (item range: 0‐3, total score: 0‐21), with higher scores indicating higher levels of depressive symptoms. Anxiety was evaluated by the Beck Anxiety Inventory (BAI) [], a 21-item 4-point Likert scale (item range: 0‐3, total score: 0‐63), with higher scores indicating higher levels of anxiety symptoms. Perceived stress was evaluated using the Perceived Stress Scale (PSS), a 10-item, 4-point Likert scale (item range: 0‐4, total score: 0‐40) []. Quality of life was evaluated by the 16-item, 7-point (item range: 1‐7, total score: 16‐112) Quality of Life Scale (QOLS) [].

Statistical Analysis

Following calculations of descriptive statistics for demographic and psychological characteristics, possible between-group differences at preintervention were evaluated using independent samples t test for numerical variables, and chi-square tests of independence for categorical variables. Missing data were minimal (ie, <10‐15% of total sample), and there were no statistically significant differences in study outcomes at preintervention between participants who completed the study and those who dropped out. To estimate missing observations, we used multiple imputations and produced 10 imputed datasets. Statistical analyses were performed using SPSS (version 26.0; IBM Corp) []. Multiple imputation was performed in RStudio (version 4.2; Posit) [], using the Multivariate Imputation by Chained Equations (MICE) [] package, and mediation analysis, using a structural equation modeling approach, was performed by lavaan [], and lavaan.mi [] packages. Lavaan.mi provides pooled parameter estimates and CIs based on multiple imputed datasets, according to Rubin rules. Monte Carlo Simulations based on 20,000 iterations were used to determine the 95% CIs.

Model Specification

Specification of the proposed model was guided by the intervention’s M-CBT theoretical perspective and evaluation of direct (ie, nonmediated) intervention effects on study outcomes. Subsequently, a conceptual model was specified (see ) to assess the role of NRE (as the sole FFMQ-SF facet with statistically significant direct effects) as a mediator of intervention effects on depression and anxiety (symptom-driven outcomes), and perceived stress and quality of life (as functional outcomes). For perceived stress, the model additionally explored depression and anxiety as mediators. For quality of life, the model included depression, anxiety, and perceived stress as mediators. The extended model (see ) included additional adjustments for other mindfulness facets at preintervention, to take into account possible individual differences and associations between FFMQ-SF facets and study outcomes. Both models specified covariances among exogenous variables (ie, preintervention levels of study outcomes). In both models, all proposed mediators and outcomes at postintervention (ie, 8 wk) were adjusted for their respective preintervention levels. Given elevated correlations between depression and anxiety at postintervention (r=.70 to r=.75 across imputed datasets), the covariance between their residual errors post intervention was additionally specified. Similarly, both depression and anxiety at postintervention were adjusted for each other’s baseline levels.

‎Figure 1. Mediation model indicating direct and indirect effects of MVC (compared to waitlist control) on study outcomes (β, *P<.05, **P<.01, ***P<.001; n= 306). BAI: Beck Anxiety Inventory; MVC Mindfulness Virtual Community; NRE: nonreactivity; PHQ-9: Patient Health Questionnaire-9; PSS: Perceived Stress Scale; QOLS: Quality of Life Scale; T1: baseline; T2: 8 weeks post intervention. ‎Figure 2. Mediation model (extended) indicating direct and indirect effects of Mindfulness Virtual Community (compared to waitlist control) on study outcomes, adjusted for preintervention Five Factor Mindfulness Questionnaire-Short Form (FFMQ-SF; β, *P<.05, **P<.01, ***P<.001; n= 306). AAW: acting with awareness; DES: describing; MVC: Mindfulness Virtual Community; NJU: nonjudgment; NRE: nonreactivity; OBS: observing; PHQ-9: Patient Health Questionnaire-9; PSS: Perceived Stress Scale; QOLS: Quality of Life Scale; T1: baseline; T2: 8 weeks post intervention. Model Fit

Model fit was evaluated according to multiple indices, including the comparative fit index (CFI), the root-mean-square error of approximation (RMSEA), and the standardized root-mean-square residual (SRMR). Models were deemed good or acceptable fit with a CFI≥.95, an RMSEA<.08, an SRMR<.08, and in line with standard recommendations [].

Ethical Considerations

This research was approved by the Human Participants Research Committee at York University (Certificate: 2023 - 012).

ResultsParticipants

Study participants were 306 undergraduate university students who were enrolled in the intervention program in 2017 (n=148) and 2018 (n=158) cohorts with complete psychometric evaluations at preintervention. Altogether, n=29 participants (9.5%) across both cohorts dropped out without completing postintervention assessments, with 18 from the 2017 and 11 from the 2018 cohorts. Although reasons for attrition were not indicated in original publications [,], attrition rate did not differ by cohort year (P=.12), and there were no statistically significant differences in demographics or psychological characteristics among the participants who dropped out and those who completed the 8-week MVC intervention.

As shown in , no statistically significant differences in demographic or psychological characteristics were detected at preintervention between MVC and WLC groups. These included mean age (P=.59), gender (P=.10), country of birth (P=.25), first language (P=.47), relationship status (P=.06), ethnic background (P=.90), PHQ-9 (P=.38), BAI (P=.80), PSS (P=.50), QOLS (P=.35), and FFMQ-SF facets (P=.17 to P=.71). In addition, when comparing preintervention demographic and psychological characteristics between participants from the years 2017 and 2018, no statistically significant differences were seen, except for higher proportions of single (no relationship) participants in the 2018 WLC group and married participants in the 2017 WLC group (P=.001).

Table 1. Demographic and psychological characteristics of study participants at preintervention.Demographic characteristicsMVC (n=152)WLC (n=154)P valueAge, mean years (SD)23.21 (7.53)23.73 (9.23).59Gender, n (%).10Male40 (26.3)25 (16.2)Female111 (73)128 (83.1)Other1 (0.7)1 (0.6)Country of birth, n (%).25Canada80 (52.6)71 (46.1)Other72 (47.4)83 (53.9)Years in Canada, mean (SD)9.8 (9.1)9.0 (7.1).56First language, n (%).47English95 (62.5)90 (58.4)Other57 (37.5)64 (41.6)Relationship status, n (%).06Single81 (53.3)96 (62.3)Single (in relationship)56 (36.8)36 (23.4)Married or common law relationship13 (8.6)17 (11)Other2 (1.3)5 (3.2)Ethnic background, n (%).90White29 (19.1)27 (17.5)Black21 (13.8)18 (11.7)South Asian37 (24.3)42 (27.3)Other49 (32.2)47 (30.5)Multiple ethnicities16 (10.5)20 (13)Self-rated health, n (%).27Poor or fair26 (17.1)36 (23.4)Good54 (35.5)57 (37)Very good or excellent72 (47.4)61 (39.6)Access to private mental health, n (%).42Yes59 (38.8)53 (34.4)No93 (61.2)101 (65.6)Weekly hours, mean hours (SD) Paid work9.32 (9.77)9.13 (11.44).87Unpaid work (incl. volunteer work)2.71 (4.76)3.45 (5.69).22Vigorous physical activities2.64 (5.51)2.22 (5.4).50Psychological Characteristicsd PHQ-9, mean (SD)8.71 (6.4)9.36 (6.49).38PHQ-9 score ranges, n (%).630‐990 (59.2)87 (56.5)≥ 1062 (40.8)67 (43.5)BAI, mean (SD)15.18 (12.1)15.51 (11.39).80BAI range, n (%).760‐21 (low)111 (73)110 (71.4)≥ 22 (moderate-high)41 (27)44 (28.6)PSS score, mean (SD)19.97 (7.78)20.56 (7.55).50PSS score ranges, n (%).760‐13 (low)28 (18.4)27 (17.5)14‐26 (moderate)91 (59.9)88 (57.1)27‐40 (high)33 (21.7)39 (25.3)QOLS, mean (SD)81.15 (15.95)79.51 (15.22).35FFMQ-SF, mean (SD) DES16.10 (4.16)16.58 (4.31).31OBS13.59 (3.43)13.88 (3.27).44AAW16.44 (4.31)16.24 (4.72).71NJU14.12 (3.9)13.95 (4.37).70NRE13.85 (3.97)14.44 (3.68).17

aMVC: Mindfulness Virtual Community.

bWLC: waitlist control.

cBased on 155 participants born outside of Canada.

dThe other category includes engaged, divorced, separated, or widowed.

eBased on 303 participants.

fPHQ-9: Patient Health Questionnaire-9.

gBAI: Beck Anxiety Inventory.

hPSS: Perceived Stress Scale.

iQOLS: Quality of Life Scale.

jFFMQ-SF: Five Facet Mindfulness Questionnaire-Short Form.

kDES: describing.

lOBS: observing.

mAAW: acting with awareness.

nNJU: nonjudgment.

oNRE: nonreactivity.

Between-Group Differences in Study Outcomes

presents means (SD) for MVC and WLC groups at pre (ie, baseline) and at 8 weeks post intervention. The effect of study group (ie, MVC vs WLC) on outcomes (ie, direct nonmediated effects) was indicated by statistically significant between-group differences in study outcomes post intervention, adjusted for baseline levels of the respective outcomes. Results indicated statistically significant differences between MVC and WLC groups post intervention on the PHQ-9 (b=−1.72; P=.002), BAI (b=−3.40; P=.001), PSS (b=−2.44; P<.001), QOLS (b=4.31; P=.005), and the NRE facet of the FFMQ-SF (b=1.63; P<.001), all favoring the MVC.

Table 2. Postintervention differences (8 weeks) in study outcomes between Mindfulness Virtual Community (MVC) and Waitlist Control (WLC) groups.Outcomes and timeMVC (n=152)WLC (n=154)B (95% CI)βP valuePHQ-9−1.72 (−2.78 to −0.64)−0.13.002Preintervention8.71 (6.40)9.36 (6.49)Postintervention7.81 (6.14)9.98 (6.82)BAI−3.40 (−5.42 to −1.37)−0.15.001Preintervention15.18 (12.10)15.51 (11.39)Postintervention11.84 (10.69)15.45 (11.86)PSS−2.44 (−3.78 to −1.10)−0.16<.001Preintervention19.97 (7.78)20.56 (7.55)Postintervention18.30 (7.03)21.11 (8.00)QOLS4.31 (1.29 to 7.32)0.13.005Preintervention81.15 (15.95)79.51 (15.22)Postintervention80.80 (15.33)75.42 (17.15)FFMQ-SFDES0.37 (−0.35 to 1.09)0.05.32Preintervention16.10 (4.16)16.58 (4.31)Postintervention16.40 (3.67)16.33 (4.34)OBS0.60 (−0.04 to 1.25)0.09.07Preintervention13.59 (3.43)13.88 (3.27)Postintervention13.74 (3.66)13.32 (3.34)AAW0.25 (−0.51 to 1.02)0.03.52Preintervention16.44 (4.31)16.24 (4.72)Postintervention15.80 (3.75)15.46 (3.93)NJU−0.07 (−0.86 to 0.71)−0.009.85Preintervention14.12 (3.90)13.95 (4.37)Postintervention14.12 (4.07)14.10 (4.26)NRE1.63 (0.89 to 2.37)0.22<.001Preintervention13.85 (3.97)14.44 (3.68)Postintervention15.71 (3.57)14.38 (3.89)

aMVC: Mindfulness Virtual Community.

bWLC: waitlist Control.

cBetween-group differences at postintervention, adjusted for each outcome’s baseline levels. b and β (unstandardized and standardized estimates respectively) indicate the direct (ie, nonmediated) effect of the MVC intervention on study outcomes.

dPHQ-9: Patient Health Questionnaire-9.

eBAI: Beck Anxiety Inventory.

fPSS: Perceived Stress Scale.

gQOLS: Quality of Life Scale.

hFFMQ-SF: Five Facet Mindfulness Questionnaire-Short Form.

iDES: Describing.

jOBS: Observing.

kAAW: acting with awarenesss.

lNJU: Nonjudgment.

mNRE: Nonreactivity.

Mediation Models

Following evaluation of post-intervention differences between MVC and WLC groups (ie, adjusted direct effects), NRE, as the sole FFMQ-SF facet impacted by the MVC intervention, was included in the mediation model. In accordance with the hypothesized model (see and ), results supported NRE as a mediator of the effects of MVC intervention on depression (b=−.41; P=.03), anxiety (b=−.93; P=.003), and perceived stress (b=−.70; P=.002), but not quality of life (b=.34; P=.32).

Table 3. Direct and indirect paths per study outcome in the main mediation model.Outcomes and pathsB (95% CI)βP valueDepressionMVC>NRE1.63 (0.86‐2.40)0.21<.001NRE>PHQ-9−0.25 (−0.42 to −0.08)−0.15.004MVC>NRE>PHQ-9−0.41 (−0.79 to −0.11)—.03AnxietyNRE>BAI−0.57 (−0.85 to −0.30)−0.20<.001MVC>NRE>BAI−0.93 (−1.67 to −0.36)—.003Perceived stressNRE>PSS−0.43 (−0.59 to −0.27)−0.22<.001MVC>NRE>PSS−0.70 (−1.18 to −0.31)—.002MVC>PHQ-9−1.44 (−2.54 to −0.34)−0.11.01PHQ-9>PSS0.45 (0.31‐0.58)0.39<.001MVC>PHQ-9>PSS−0.64 (−1.24 to −0.14)—.02MVC>NRE>PHQ-9>PSS−0.18 (−0.37 to −0.05)—.03MVC>BAI−2.48 (−4.44 to −0.52)−0.11.01BAI>PSS0.07 (0.003‐0.14)0.11.04MVC>BAI>PSS−0.18 (−0.45 to 0.001)—.11MVC>NRE>BAI>PSS−0.07 (−0.17 to −0.003)—.09Quality of lifeNRE>QOLS0.21 (−0.18 to 0.59)0.05.29MVC>NRE>QOLS0.34 (−0.29 to 1.04)—.32PHQ-9>QOLS−0.10 (−0.48 to 0.29)−0.04.62MVC>PHQ-9>QOLS0.14 (−0.47 to 0.76)—.61MVC>NRE>PHQ-9>QOLS0.04 (−0.12 to 0.24)—.68BAI>QOLS−0.23 (−0.44 to −0.02)−0.16.03MVC>BAI>QOLS0.56 (0.002‐1.50)—.05MVC>NRE>BAI>QOLS0.21 (0.01‐0.53)—.03MVC>PSS−0.83 (−1.96 to 0.31)-0.06.15PSS>QOLS−0.65 (−0.91 to −0.38)-0.30<.001MVC>PSS>QOLS0.54 (−0.20 to 1.35)—.21MVC>NRE>PSS>QOLS0.45 (0.17‐0.85)—.02MVC>PHQ-9>PSS>QOLS0.41 (0.08‐0.93)—.05MVC>BAI>PSS>QOLS0.12 (−0.001 to 0.32)—.16MVC>NRE>PHQ-9>PSS>QOLS0.12 (0.03‐0.27)—.08MVC>NRE>BAI>PSS>QOLS0.04 (0.002‐0.12)—.14MVC>QOLS1.54 (−1.00 to 4.07)0.05.23

aMVC: Mindfulness Virtual Community.

bNRE: Nonreactivity.

cPHQ-9: Patient Health Questionnaire-9.

dNot available.

eBAI: Beck Anxiety Inventory.

fPSS: Perceived Stress Scale.

gQOLS: Quality of Life Scale.

For perceived stress, results additionally supported the mediating role of depression (b=−.64; P=.02), but not anxiety (b=−.18; P=.11). For quality of life, results indicated mediation through anxiety (b=.57; P=.05), but not depression (b=.14; P=.61). The effects of MVC intervention on PSS were further mediated by a sequential mediation path through NRE and PHQ (b=−.18; P=.03). Similarly, quality of life was mediated by sequential mediation paths through NRE and BAI (b=.21; P=.03), and NRE and PSS (b=.45; P=.02; see ).

Interestingly, in the presence of mediated (ie, indirect) effects, the direct paths from the MVC intervention to depression (b=−1.44; P=.01) and anxiety (b=−2.48; P=.01) reduced in magnitude when compared to direct effect models (see ). Similarly, direct paths from the MVC intervention to perceived stress (b=−0.83; P=.15) and quality of life (b=1.54; P=.23) were no longer statistically significant. Model fit indices indicated acceptable fitness (CFI=.988; RMSEA=.052; SRMR=.05). The model accounted for 52.1% of variability in depression, 49.3% of variability in anxiety, 64.5% of variability in perceived stress, and 57.5% of variability in quality of life.

In the extended model (see and ), all outcomes were further adjusted for preintervention levels of mindfulness (ie, FFMQ-SF) facets. Given that mindfulness is both regarded as an individual disposition (ie, trait) and a cultivated capacity, an extension of the model additionally accounted for individual differences in dispositional mindfulness. Overall, results from the extended model (see and ) were similar to the main model (see and ). However, preintervention FFMQ-SF facets showed different patterns of association with study outcomes. Specifically, among the 5 mindfulness facets at preintervention, statistically significant relationships were observed between DES and depression (b=.19; P=.02), OBS and anxiety (b=.30; P=.05), NJU and perceived stress (b=−.23; P=.001), NJU and quality of life (b=−.37; P=.03). Model fit indices for the extended model indicated good fitness (CFI=.992; RMSEA=.043; SRMR=.036).

Table 4. Direct and indirect paths per study outcome in the extended mediation model.Outcomes and pathsB (95% CI)βP valueDepressionMVC>NRE1.63 (0.87‐2.39)0.21<.001NRE>PHQ-9−0.30 (−0.49 to −0.13)−0.17.004MVC>NRE>PHQ-9−0.48 (−0.91 to −0.13)—.02AnxietyNRE>BAI−0.63 (−0.93 to −0.33)−0.21<.001MVC>NRE>BAI−1.02 (−1.80 to −0.41)—.002Perceived stressNRE>PSS−0.48 (−0.64 to −0.32)−0.25<.001MVC>NRE>PSS−0.78 (−1.30 to −0.36)—.001MVC>PHQ-9−1.22 (−2.37 to −0.06)−0.09.04PHQ-9>PSS0.44 (0.30‐0.57)0.38<.001MVC>PHQ-9>PSS−0.53 (−1.13 to −0.02)—.04MVC>NRE>PHQ-9>PSS−0.21 (−0.43 to −0.05)—.03MVC>BAI−2.28 (−4.23 to −0.32)−0.10.02BAI>PSS0.08 (0.008‐0.15)0.12.03MVC>BAI>PSS−0.18 (−0.44 to −0.001)—.13MVC>NRE>BAI>PSS−0.08 (−0.19 to −0.008)—.08Quality of lifeNRE>QOLS0.34 (−0.09 to 0.76)0.08.12MVC>NRE>QOLS0.55 (−0.13 to 1.40)—.17PHQ-9>QOLS−0.12 (−0.49 to 0.25)−0.05.52MVC>PHQ-9>QOLS0.15 (−0.35 to 0.72)—.53MVC>NRE>PHQ-9>QOLS0.06 (−0.12 to 0.29)—.60BAI>QOLS−0.21 (−0.41 to −0.007)−0.15.04MVC>BAI>QOLS0.48 (−0.01 to 1.37)—.06MVC>NRE>BAI>QOLS0.22 (0.005‐0.53)—.03MVC>PSS−0.73 (−1.84 to 0.38)−0.05.20PSS>QOLS−0.70 (−0.97 to −0.43)−0.32<.001MVC>PSS>QOLS0.51 (−0.27 to 1.36)—.24MVC>NRE>PSS>QOLS0.55 (0.23‐0.99)—.007MVC>PHQ-9>PSS>QOLS0.37 (0.01‐0.89)—.08MVC>BAI>PSS>QOLS0.13 (0.00‐0.34)—.18MVC>NRE>PHQ-9>PSS>QOLS0.15 (0.03‐0.33)—.06MVC>NRE>BAI>PSS>QOLS0.06 (0.005‐0.14)—.12MVC>QOLS1.06 (−1.52 to 3.65)0.03.42

aMVC: Mindfulness Virtual Community

bNRE: nonreactivity.

cPHQ-9: Patient Health Questionnaire-9.

dNot available.

eBAI: Beck Anxiety Inventory.

fPSS: Perceived Stress Scale.

gQOLS: Quality of Life Scale.

DiscussionPrincipal Findings

In this study we evaluated a mediation model for the MVC intervention, aimed at clarifying its therapeutic mechanisms, based on data from 2 RCTs involving undergraduates at a Canadian university. The mediation model (see and ) supported the role of the NRE facet of FFMQ-SF as a mediator of the MVC intervention effects on depression, anxiety, perceived stress, and quality of life outcomes. Our results further supported the role of depression as a mediator for the MVC intervention effects on perceived stress, alongside other sequential mediation paths for perceived stress and quality of life. Changes in NRE reflect an important intervention effect, indicating participants’ increased capacities to discontinue automatic depressogenic-anxiogenic reactions to events. This is exemplified in the mindfulness of breath technique emphasized, where participants were instructed to direct awareness to breathing sensations and return attention to them after thoughts and emotions arise, creating a “gap” between stimuli exposure and automatic reaction. This gap introduces a context for altering automatic responding, resulting in responses that reflected increased awareness and decreased avoidance. Therefore, phenomena that may have previously been viewed automatically (in a negative frame) could now be re-perceived in open-ended, neutral, and potentially positive ways.

According to Burzler and Tran’s [] proposed model of mindfulness development with respect to the five facets model, mindfulness cultivation involves progressive developments in OBS, DES, and AAW facets, indicating enhanced attentional deployment, which then stimulates further development of the NJU and NRE facets. Burzler and Tran’s [] model emphasizes development in the attentional elements of the construct (ie, OBS, DES, and AAW) before development of the attitudinal elements (ie, NJU and NRE). While our findings were similar to their proposed model in supporting NRE as the final outcome of mindfulness training, a comprehensive evaluation of this model requires determination of change trajectories of individual facets over the course of intervention with multiple measurements in closer succession (eg daily or weekly evaluation). In contrast to this model, there is also evidence that in mindfulness development, different attentional and attitudinal facets may interact to produce healthy outcomes. For example, the OBS facet has been shown to be associated with better sleep health at higher NRE levels but not at lower NRE levels []. Our findings emphasize the NRE facet as the key mediator of the MVC intervention. This is aligned with previous evaluations that identified NRE as a mediator for improvements in cognitive flexibility in mindfulness-based stress reduction training [], and reductions in health anxiety in cognitive-behavioral and acceptance and commitment therapies [,]. Further support for the importance of NRE to well-being and distress emerged in a study on the relative importance of FFMQ facets as dispositions in a general population sample []. Specifically, in that study, NRE accounted for the highest proportion of variance (25%) in well-being, and the second highest proportion of variance (7%) in distress after AAW (20%) [].

In contrast to signifying passivity, we suggest NRE, as conceptualized in the FFMQ-SF, reflects an attitude of nonavoidance with equanimity, defined as even-minded responsiveness to mental phenomena, unswayed by associated emotional valences []. Indeed, NRE has shown the highest magnitude of correlation with the even-mindedness subscale of the EQUA-S equanimity scale (.54; P<.001), among other FFMQ facets (AAW: .22; P<.001, NJU: .30; P<.001) []. NRE has further shown a moderate negative correlation with experiential avoidance (−.39; P<.001), alongside other FFMQ facets (DES: −.23, P<.001; AAW: −.30, P<.001; NJU: −.49, P<.001) []. In the same study, both NRE (.53, P<.001) and NJU (.48, P<.001) demonstrated the highest correlations with self-compassion [].

Strengths and Limitations

There are advantages and limitations to this investigation. First, our study used a large sample size based on data aggregated from two RCTs with identical recruitment and administration procedures. In addition, among previous evaluations of specific mindfulness facets as mediating mechanisms of MBIs [-,], our research provides key evidence on the role of NRE as a mediator of MBI with an M-CBT perspective. Second, we used a structural equation modeling approach to evaluate mediation, which, in comparison to regression-based methods, permits simultaneous evaluations of multiple inter-relationships of varying complexities (ie, parallel and sequential indirect paths). In this approach, as was the case for depression, anxiety, and perceived stress in this study, variables can serve as both mediators and outcomes within a hypothesized matrix of associations. However, as the MVC intervention evaluated outcomes at 2 time points, absence of temporal precedence of mediators in this study precludes a definite indication of causality []. Longitudinal evaluations with multiple time points are better suited to establish the temporal precedence of change in mediators before change in outcomes.

The mediation models in this study dem