Protocol for a mechanistic study of Mindfulness Based Cognitive Therapy during pregnancy

Abstract

Objective:

Preventive interventions for postpartum depression (PPD) are critical for women at elevated risk of PPD. Mindfulness Based Cognitive Therapy – Perinatal Depression (MBCT-PD) is a preventive intervention that has been shown to reduce risk for PPD in women with a prior history of depression. The objective of this clinical trial is to examine two potential mechanisms of action of MBCT-PD, emotion regulation and cognitive control, using behavioral and neuroimaging methods.

Method:

This baseline protocol describes a randomized control trial (RCT) with two arms, MBCT-PD and treatment as usual (TAU). We plan on enrolling 74 females with a prior history of a major depressive episode, with 37 participants randomized to each arm. Participants in the MBCT-PD arm will receive MBCT-PD during pregnancy, and the TAU group will receive standard prenatal care. All participants will complete the Center for Epidemiologic Studies Depression – Revised (CESD-R), Emotion Regulation Questionnaire (ERQ), and classic Stroop task at multiple points from pregnancy through six months postpartum. Participants will also complete an fMRI scan at six weeks postpartum.

Results:

All primary outcomes are collected at six weeks postpartum. Primary behavioral outcomes include: depressive symptoms on the CESD-R, cognitive reappraisal on the ERQ, and Stroop task performance. In parallel, the primary neurobiological outcomes include whole-brain activation during fMRI tasks when participants 1) regulate emotional responding and 2) engage cognitive control.

Conclusions:

This results of this innovative RCT will help identify potential behavioral and neurobiological mechanisms of action of preventive interventions for PPD for in-depth examination in larger scale RCTs.

Study Rationale

Background

Approximately 10–15% of women will experience postpartum depression (PPD), making it one of the most common complications postpartum (Gavin et al., 2005). PPD often emerges early in the postpartum period, with peak presentation at six weeks postpartum (Gavin et al., 2005). Approximately 30–40% of women with a prior history of depression will experience PPD (Goodman & Tully, 2009). PPD is a pressing concern not only because of the effects it can have on maternal well-being, but also the concurrent and prospective effects it can have on offspring. PPD can be associated with decreased maternal responsiveness, poorer infant attachment, and greater rates of behavioral and clinical difficulties in children (Stein et al., 2014).

The perinatal period is marked by enhanced neuroplasticity, particularly in brain regions that support detection of salient information, emotional reactivity, executive function, and empathic responding (Barba-Müller, Craddock, Carmona, & Hoekzema, 2019; Kim, Strathearn, & Swain, 2016). Structural and functional changes to the brain likely support maternal responsiveness (Kim et al., 2016) and may be related to changes in cognitive and emotional processes during the perinatal period, including mild cognitive deficits (e.g., working memory) and increased vigilance to threat, that are thought to promote protection of the fetus/child and prioritization of cognitive resources for actions needed to prepare (prenatal) and care (postpartum) for the infant (Anderson & Rutherford, 2012; Barba-Müller et al., 2019). However, women with PPD demonstrate altered activation of brain regions and connectivity implicated in emotional reactivity and executive function (Pawluski, Lonstein, & Fleming, 2017).

Pharmacological interventions are often recommended for women at-risk for PPD, but most women prefer behavioral interventions during pregnancy (Dimidjian & Goodman, 2014) and over 50% of pregnant women discontinue use of anti-depressants (Yonkers, Blackwell, Glover, & Forray, 2014). Recognizing the need for preventive interventions for PPD, the U.S. Preventive Services Task Force (USPSTF) recommends all pregnant and postpartum women at-risk for depression be referred for behavioral intervention (U.S. Preventive Services Task Force, 2019). However, few interventions have been validated with high quality studies (O’Connor, Senger, Henninger, Coppola, & Gaynes, 2019). Mindfulness Based Cognitive Therapy-Perinatal Depression (MBCT-PD) is one of the few effective preventive interventions for reducing risk of PPD in women with a prior history of depression (Dimidjian et al., 2015, 2016; O’Connor et al., 2019). MBCT-PD is an adaptation of Mindfulness Based Cognitive Therapy (MBCT) for the perinatal period (Dimidjian et al., 2015, 2016). MBCT is a preventive group intervention that promotes awareness, distress tolerance, acceptance, and coping skills for managing thoughts and behaviors that increase susceptibility to depression (Segal, Williams, & Teasdale, 2013). There is strong evidence that MBCT is effective in reducing risk for future depressive episodes (Chiesa & Serretti, 2011; Piet & Hougaard, 2011), including PPD (Dimidjian et al., 2015, 2016).

A critical next step is identifying mechanisms through which MBCT-PD reduces risk of PPD. A meta-analysis of mediation studies of MBCT has found evidence that, as expected, mindfulness is a mediator of the effect of MBCT on levels of depression (Gu, Strauss, Bond, & Cavanaugh, 2015). The same meta-analysis found strong evidence for emotion regulation as a mechanism of action of MBCT (Gu et al., 2015), and other studies have suggested cognitive control may also be a mechanism of action (van der Velden et al., 2015). Emotion regulation is the active management of a strong emotional response (Silvers et al., 2015), and cognitive control is the ability to sustain attention for task relevant information in the face of distracting information (Banich, 2009). Neuroimaging studies further support emotion regulation and cognitive control as promising mechanisms of action of MBCT (Chiesa, Serretti, & Jakobsen, 2013). Theoretically, mindfulness is a tool that may increase emotion regulation and cognitive control, two potential mechanisms of action of MBCT, by increasing distress tolerance, awareness, and the ability to implement cognitive and behavioral skills.

Study Objectives

This is a baseline protocol for a pilot clinical trial that employs an experimental medicine approach, consistent with the Science of Behavior Change (SOBC) program, to examine candidate mechanisms of action of MBCT-PD. The primary objective of the study is to determine whether emotion regulation is a candidate behavioral mechanism of action of MBCT-PD (see Figure 1A). Specifically, the study examines whether MBCT-PD increases emotion regulation and decreases depressive symptoms postpartum. One of the secondary objectives is to determine whether neuroimaging data supports emotion regulation as a candidate mechanism of action of MBCT-PD. We hypothesize that individuals who complete MBCT-PD during pregnancy will report greater use of re-appraisal, an effective emotion regulation strategy that involves modifying the intensity of an emotional experience, and will concomitantly demonstrate greater activation of ventral medial prefrontal brain regions that down regulate emotional reactivity six weeks postpartum than controls. Additional secondary objectives include examining whether cognitive control is a candidate mechanism of action of MBCT-PD using both behavioral and neuroimaging data (see Figure 1B). We predict that individuals who complete MBCT-PD during pregnancy will demonstrate better performance on a cognitive control task as well as greater activation of lateral prefrontal brain regions important for cognitive control at six weeks postpartum compared to controls.

Figure 1.

Proposed mechanisms of action of MBCT-PD: emotion regulation (A) and cognitive control (B).

Study Design

The study is a randomized control trial with an intervention arm, MBCT-PD, and control arm, treatment as usual (TAU) with a delayed postpartum mindfulness psychoeducation session. TAU was chosen as a control arm because no mechanistic studies have been performed to date with MBCT-PD. We plan to enroll 74 pregnant females with a prior history of a major depressive episode (MDE), with 37 participants randomized to each arm. Primary behavioral and neurobiological outcomes will be assessed at six weeks postpartum.

Methods

Participants

Recruitment.

Primary recruitment will occur through Oregon Health & Science University (OHSU), including recruitment through the obstetric and midwifery services and family medicine clinics, review of electronic medical records, and referrals from other IRB-approved prenatal studies. Additional recruitment from the surrounding urban community will also be implemented via local advertisements and social media.

Sample.

Enrollment.

Full inclusion and exclusion criteria will be assessed prior to enrollment. Criteria related to psychiatric diagnoses will be assessed by the study team. Primary inclusion criteria includes: pregnant (biological sex is female); single gestation; 14–22 weeks gestational age (GA); age 21–45; prior history of at least one MDE; available and physically able to attend MBCT-PD sessions; and able to understand and comply with study procedures. Participants need to be fluent in English, as the MBCT-PD protocol has not been validated in any other languages.

Exclusion criteria include: MRI contraindication; current MDE or manic/hypomanic episode; active suicidality; use of psychotropic medications or other CNS medication; current substance use disorder (except early remission) or any illicit substance use; Intellectual Disability Disorder or Autism Spectrum Disorder; lifetime history of psychosis; significant head injury; prior history of engaging in MBCT, mindfulness-based stress reduction (MBSR), or the mindfulness module of dialectical behavior therapy; major neurological or medical illness; known congenital, genetic, or neurologic disorder of the fetus; and IQ less than 80.

Participants will remain active in the study if they initiate psychotherapeutic intervention or psychotropic medication after enrollment. Participants will not undergo MRI scanning if they develop a MRI contraindication after enrollment or are discontinued from the intervention. With the exception of participants discontinued due to miscarriage, still birth, or abortion, participants will continue to be followed with their permission.

Consent.

Participants will provide digital consent prior to completion of the initial electronic screen, and full informed, written consent (Appendix) at the time of their initial in-person visit. The consent stipulates that there is no financial compensation for perceived or actual harm. The study protocol is approved by the Institutional Review Board (IRB) at OHSU and the National Center for Complementary and Integrative Health (NCCIH). All amendments and protocol changes are submitted to the IRB for approval, approved directly by the NCCIH, and updated as appropriate on ClinicalTrials.gov. We have also obtained a Certificate of Confidentiality from the National Institutes of Health (NIH).

Interventions

Mindfulness Based Cognitive Therapy – Perinatal Depression (MBCT-PD).

Each MBCT-PD group will include three to eight participants who will attend eight sequential, weekly 2-hour group sessions co-led by two master’s level psychotherapists. MBCT-PD includes mindfulness and CBT skills. The mindfulness skills are designed to increase capacity for present moment awareness, which is a precursor for the core elements of CBT: identifying and changing thoughts and behaviors (Dimidjian et al., 2015, 2016; Segal et al., 2013). All sessions start with a mindfulness exercise and inquiry, and then a review of at-home practice. The rest of the session includes psychoeducation, introduction of new skills, and additional mindfulness exercises. Participants are assigned at-home practices for the coming week and audio recordings are provided to guide all at-home formal mindfulness practices. All sessions are video-recorded for later fidelity assessment. Individuals in this arm will receive standard prenatal care as well.

Participants will be discontinued from the intervention if the following occur: active suicidality; serious mental illness; substantial substance use; initiation of any inpatient, sub-acute, or intensive outpatient psychiatric services; or in the event of a miscarriage, still birth, or abortion. In the event that a participant develops a new pregnancy complication, they will be allowed to actively continue as long as their medical provider documents that it is safe.

Treatment as usual. (TAU).

Participants in the TAU group will receive routine prenatal care from a prenatal provider of their choosing. They will be offered the option of attending a 2-hour group mindfulness psychoeducation session at six to nine months postpartum (after their last postpartum visit). The mindfulness psychoeducation session introduces the concept of mindfulness, some brief mindfulness activities, and basic CBT principles. Individuals will also receive a copy of the MBCT-PD workbook (Dimidjian & Goodman, 2019), which includes the audio recordings.

Concomitant interventions

Due to ethical concerns, there are no prohibited behavioral or medical interventions for participants in either arm of the study. Ancillary medical or psychiatric treatments, psychotherapy, and mindfulness-based activities (e.g., yoga) will be regularly tracked.

Randomization

Participants will be randomized using a blocked design with blocks of two to four. The study coordinator will enroll and randomize eligible participants by executing a REDCap module pre-programmed by the study statistician’s staff. To create the randomization script, a data frame with the number of rows corresponding to the randomly selected block size will be created with a balanced number of assignments to each arm. Each assignment will be given a random number from a uniform distribution and then ordered by their number to randomly shuffle the study arm assignments within each block.

Study Flow

See Figure 2 for study timeline. Gross eligibility (e.g., MRI contraindications) will be assessed via an electronic screen and full eligibility at an in-person eligibility visit (14–22 weeks GA). Individuals deemed eligible for the full study after a consensus meeting will be randomized within two weeks of their visit. Participants randomized to MBCT-PD will complete pre-treatment, mid-treatment, and post-treatment surveys electronically. Participants in the TAU arm will complete the surveys at the same time intervals. All participants will complete in-person visits at 34 weeks GA, six weeks postpartum (including MRI), and six months postpartum.

Figure 2.

Timeline of study visits and data collection point. N’s reflect expected participants at each time point, accounting for expected attrition. GA = gestational age. wk(s) = week(s). mos = months.

Masking

Participants and interventionists cannot be masked. The study coordinators are the only other study members who will not be masked. Data will be tracked in REDCap, and group will be coded with a proxy (e.g., A and B) for analysis purposes. The mask will be removed by the study statistician once analyses of the primary outcomes have been conducted.

Data Collection

See Table 1 for measurement schedule. Most survey measures will be completed via REDCap, a HIPAA compliant electronic direct data capture system. All data are time stamped and locked.

Table 1.

Primary Study Measures and Administration Time Points

Measures	Eligibility Visit	Pre-Treatment	Mid – Treatment	Post –Treatment	34 Week GA Visit	6 Week PP Visit	6 Month PP Visit
Demographics
Demographic Survey	X					X	X
Psychiatric Symptoms and History
Structured Clinical Interview for DSM-5 Disorders Clinician Version (SCID-5-CV)	X				X	X	X
Psychiatric Treatment Survey	X	X	X	X	X	X	X
Center for Epidemiologic Studies Depression Scale- Revised (CESD-R)	X	X	X	X	X	X	X
Medical History
Current Medical Health and Treatment	X	X	X	X	X	X	X
Reproductive and Obstetric History	X
Medication and supplement use	X	X	X	X	X	X	X
Emotion Regulation
Emotion Regulation Questionnaire (ERQ)a	X			X	X	X	X
Emotion Regulation Task (fMRI)						X
Cognitive Assessment
Wechsler Abbreviated Scale of Intelligence, Second Edition (WASI-II)	X
Stroop Task (Behavioral)	X					X	X
Stroop Task (fMRI)						X
Mindfulness Skills/Activities
Five Facet Mindfulness Questionnaire (FFMQ) a	X	X	X	X	X	X	X
Mindfulness Activities or Treatment	X	X	X	X	X	X	X
Log of Assigned MBCT-PD Practice			X	X	X
Stress
Perceived Stress Scale (PSS)	X	X	X	X	X	X	X

Notes: GA = gestational age. PP = postpartum.

^a= SOBC Repository measure.

Enrollment criteria and baseline characteristics.

Demographics.

Demographic information collected from all participants includes: race, ethnicity, income, education, job, food security, housing, and partner information. Updated demographic information and demographic information on the infant will be collected at the postpartum visits.

Medical and psychiatric history.

Medical and psychiatric history will be collected using surveys and medical record review. Information on all prior and current psychiatric treatment, interventions, and diagnoses as well as current psychotropic medication will be collected via survey at the eligibility visit. Medication, supplement/vitamin, and substance use will be collected via survey, including for the three months prior to pregnancy. Current medical diagnoses, pregnancy complications, and prenatal care will be reported at the eligibility visit; changes to psychiatric or medical history, diagnoses, medications, interventions, or perinatal care will be obtained at each time point.

Psychiatric disorders.

Current DSM-5 (American Psychiatric Association, 2013) disorders will be assessed with the Structured Clinical Interview for DSM-V Disorders Clinician Version (SCID-5-CV) (First, Williams, Karg, & Spitzer, 2016). The SCID-5-CV will be administered by a master’s level therapist who has been trained to standard by the principal investigator (PI). All administrations will be video recorded for reliability rating purposes. The SCID-5-CV will be used to determine whether a participant has a prior history of at least one MDE and some exclusionary criteria.

IQ.

IQ will be briefly assessed using the Wechsler Abbreviated Scale of Intelligence, 2^nd Edition (WASI-II), 2-subtest version (Wechsler & Hsiao-pin, 2011).

Mindfulness.

The Five Facet Mindfulness Questionnaire (FFMQ) (Baer, Smith, Hopkins, Krietemeyer, & Toney, 2006) will be used to assess an individual’s tendency to use mindfulness. The FFMQ has good internal and divergent validity and internal consistency (Christopher, Neuser, Michael, & Baitmangalkar, 2012) and is sensitive to detecting changes from pre- to post-treatment for MBCT (Gu et al., 2016).

Stress.

Current level of personal perceived stress will be assessed with the Perceived Stress Scale (PSS) (Cohen, Kamarck, & Mermelstein, 1983), a widely used measure of stress that measures the perceived stress of daily situations over the past month. Within psychiatric populations, the PSS has demonstrated good validity and internal consistency (Hewitt, Flett, & Mosher, 1992).

Adverse events (AE).

Self-report of potential AEs, such as a new illness or diagnosis, will be reported at each time point. Psychiatric AEs will be reviewed by the PI, and an obstetrician will review all medical AEs. Suicidality will be monitored at each time point with the CESD-R, which contains an item about suicidal ideation. Current psychiatric diagnoses and suicidality will also be assessed at each in-person visit. Strict protocols are in place to ensure proper follow-up of AEs.

Primary outcomes: behavioral.

Depressive symptoms.

Depressive symptoms will be assessed using the Center for Epidemiological Studies Depression Scale - Revised (CESD-R) (Van Dam & Earleywine, 2011). The CESD-R is a standard measure of depression in community populations, demonstrating high internal consistency and both convergent and divergent validity (Van Dam & Earleywine, 2011). Total score on the CESD-R at six weeks postpartum will be the primary measure of PPD.

Emotion regulation.

Emotion regulation abilities will be assessed using the Emotion Regulation Questionnaire (ERQ) (Gross & John, 2003), an SOBC Repository measure. The ERQ is a self-report measure that assesses use of cognitive reappraisal and expressive suppression strategies. It has sound psychometric properties, including good internal consistency, convergent and divergent validity, and replication of its two factor structure (Ioannidis & Siegling, 2015). The primary outcome variable is the cognitive reappraisal sub-scale score at six weeks postpartum.

Cognitive control.

Cognitive control will be assessed using a manual response version of the classic Stroop task (Stroop, 1935). The Stroop task is a standard measure of cognitive control that is robust to practice effects, reducing the likelihood of repeated measurement drift. During the Stroop task, individuals must name the ink color of a word while ignoring the word. The words are either a color word in a different, or incongruent, color (e.g., “red” in blue ink) or a neutral, non-color word. The incongruent condition requires the most cognitive control, as there is direct conflict in processing between the ink and the word (Banich, 2009). An interference score is calculated, which reflects a percentage increase in reaction time (RT) for correct incongruent trials as compared to RT for correct neutral trials ((incongruent RT – neutral RT)/neutral RT).

Primary neurobiological outcomes.

MRI scan procedures.

Participants will be trained on all functional tasks prior to scanning. In the scanner, the participant’s head will be stabilized with foam pillows, and they will be provided with earplugs and headphones to reduce scanner noise. For all fMRI tasks, stimuli are back-projected with a mirror mounted on the head coil and an opto-isolated button box is used for responding.

MRI data acquisition.

Data will be acquired on a Siemens 3 Tesla Prisma scanner (Siemens Medical Solutions, Erlangen, Germany) at OHSU’s Advanced Imaging Research Center with a Siemens 32-channel head coil. A high-resolution, T1-weighted anatomical scan will be collected for alignment and normalization of functional images (176 slices, 1mm isotropic, matrix = 256×256, TR/TE/TI = 2500/2.88/1060 ms, flip angle = 8°, pixel bandwidth = 240Hz). Volumes will be acquired with Volumetric Navigators (vNAVS) to correct for intra-scan motion. Functional scans will be acquired using a T2*-weighted echoplaner sequence (2.4mm isotropic, matrix = 90×90, TR/TE=800/30 ms, flip angle = 52°, field of view = 216 mm, slices = 60). A pair of field maps, with the phase encoding direction reversed on one, will be acquired in order to reduce distortion.

Emotion Regulation.

The emotion regulation task is adapted from prior fMRI studies (Silvers, Wager, Weber, & Ochsner, 2015; Zimmermann et al., 2017). During the task, participants will view negative and neutral images individually and are instructed to either passively view the image (negative and neutral) or regulate their emotional response using a distancing technique (cognitive reappraisal; negative images only). Participants will be instructed on how they can distance (e.g., looking at the image as if they were not personally connected) prior to scanning. The paradigm includes presentation of a task word cue (2 seconds), followed by a fixation, then presentation of a negative or neutral image (8 seconds), then another fixation, and lastly a valence rating scale (2 seconds). Fixation length is jittered and accounts for approximately half of the trials. Negative images will be randomly assigned to condition. 25 trials of each condition will be presented over four runs. Images have been selected from the International Affective Picture System (IAPS; National Institute of Health, Center for the Study of Emotion and Attention, University of Florida) using female norms. In accordance with prior research, negative images with moderate ratings have been selected to reduce the chance of ceiling effects on the regulation trials (Silvers et al., 2015; Zimmermann et al., 2017). Disgust images, which may differentially invoke the salience network (Klucken et al., 2012), and images with infants/young children are excluded. Each image will be rated on valence and arousal after scanning using the Self-Assessment Manikin (SAM; Bradley & Lang, 1994). The primary outcome is clusters of significantly greater/less brain activation during cognitive reappraisal compared to passive viewing of negative images at six weeks postpartum.

Cognitive Control.

A hybrid blocked/event-related paradigm will be used to assess sustained cognitive control (Banich et al., 2009). Each block consists of block-specific trials (incongruent, neutral, or congruent) and a set of neutral frequent trials that occur in all blocks. Participants will respond via a button press to the ink color of each word. Words across type are matched in length. There will be 12 Stroop blocks (four repetitions of each block type), each comprised of eight target and eight neutral trials randomly distributed, and 11 blocks of interspersed fixation. All blocks are 32 seconds. The primary outcome derived from this task is clusters of significantly greater/less brain activation during the incongruent blocks compared to the neutral blocks at six weeks postpartum.

Data Analysis

Interim Analyses and Stopping Rules

No interim analyses are planned prior to collection of the primary outcomes. The study will be stopped if MBCT-PD is associated with adverse effects that call into question the safety of the intervention, based on review by the PI and IMC, difficulty in study recruitment or retention will significantly impact the ability to evaluate the primary outcomes, or new information becomes available (e.g., published trials) that necessitates stopping the trial.

Primary Behavioral Outcomes

Primary behavioral analyses will be conducted on the intent-to-treat population, defined as all participants in the enrolled (all participants who have been randomized) population with completed baseline and pre-treatment assessments. Three analyses of covariance (ANCOVAs) will be conducted to assess the magnitude of difference between MBCT-PD and TAU groups at six weeks postpartum on: total depressive symptoms on the CESD-R, Stroop interference score, and cognitive reappraisal score on the ERQ, respectively. Since MBCT-PD is a group intervention, treatment group will be coded as a multi-category variable and the averaged effect across the MBCT-PD groups will be compared to the TAU group. We have chosen this coding scheme because it allows us to quantify an overall treatment effect while controlling for group-level factors unique to each therapy group. All analyses of behavioral outcomes will be performed with α = 0.05 and will not be corrected for multiple comparisons.

Depressive symptoms (CESD-R score) and stress level (PSS score) at baseline will be included as covariates added as independent variables to the model. We will also include 1) depressive symptoms at pre-treatment, 2) Stroop interference score at baseline, or 3) cognitive reappraisal score on the ERQ at baseline, respectively for each ANCOVA. We will assess the impact of including two other covariates, GA and baseline mindfulness (FFMQ score), to the models. We will determine whether they change the treatment effect by ≥ 5%. If so, we will present treatment effects from the model’s fit within subgroups of GA and FFMQ. Robust variance estimation, using the Huber-White sandwich estimator (White, 1980), will be used to estimate confidence intervals.

Mixed effects modeling with the above three outcome measures as dependent variables will be used as sensitivity analyses. Independent variables/covariates will be the same as ANCOVA models. Random effects will contain parameters that account for correlation within groups in the intervention arm, but allow all the participants in the control arm to be independent.

For cognitive control and emotion regulation indicators showing both a significant intervention effect and a statistically significant association with depressive symptom change from pre-treatment to six weeks postpartum, we will test for mediation with a separate mediation model for each indicator. We will use a model with CESD-R score as the dependent variable and add the six week postpartum emotion regulation or cognitive control measurement along with the treatment group and covariates, respectively. The attenuation of the treatment group effect in this model versus the one for the primary outcomes will represent degree of mediating effects. In a more formal mediation analysis, we will also examine the indirect path (as in structural equation modeling framework) from the intervention to post-treatment depressive symptoms via the emotion regulation or cognitive control indicator based on bias-corrected confidence intervals generated with bootstrap sampling. In this model, CESD-R score at baseline, PSS score at baseline, pre-treatment CESD-R score, and GA at intake will again be included as covariates.

Missing Data

In order to promote retention and reduce missing data: REDCap will be used for remote data collection to reduce visit burden; payment will be provided per assessment completed to incentivize completion of each individual set of assessments; and assistance with childcare and transportation will be provided. Suggestions for handling missing data as outlined by Jakobsen and colleagues (Jakobsen, Gluud, Wetterslev, & Winkel, 2017) will be followed. Complete cases analysis will be performed if missingness is negligible (≤5% across all groups) and unrelated to treatment group and/or pre-treatment depressive score. If missingness is between 5% and 25% in the primary outcome variable(s) only, then complete case analyses will be performed; however, “best” and “worst” case scenarios for missing data to estimate plausible range of MBCT vs. TAU effects will be imputed. If missing data is <25% in primary outcome variable(s) and missingness is related to treatment group and/or pre-treatment depressive score, then a Markov Chain Monte Carlo (MCMC) multiple imputation method will be used to impute missing values for primary outcomes, and analyze accordingly across five imputed datasets.

Primary Neurobiological Outcomes

fMRI pre-processing.

A modified version of the original Human Connectome Project pipeline (Glasser et al., 2013), the freely available ABCD-HCP pipeline (https://github.com/DCAN-Labs/abcd-hcp-pipelines) described in Marek et al., 2019, will be used to process and analyze fMRI and anatomical data. The pipeline is comprised of the following stages: 1) PreFreesurfer for bias field correction, brain extraction and registration of anatomical data to the Montreal Neurological Institute stereotactic space (MNI152), with ANTS denoising and bias field correction; 2) Freesurfer for automatic segmentation of brain structures, surface generation in native space, and surface registration to a standard template; 3) Postfreesurfer for conversion to CIFTI format and transforming volumes to standard space; 4) Vol for bias correction and registration of functional data to standard space via the structural data; and 5) Surf to project functional data to the template surface, including resampling of fMRI volume data first to the individual native surface and then to the template surface (32k fs_LR surface) in a single resampling step and combining CIFTI formatted volumetric subcortical and surface functional data with Connectome Workbench (Marcus et al., 2011). Frame-frame realignment is also performed, generating framewise displacement (FD) measures (calculated as the square sum of all the motion vectors estimated during frame-frame alignment). Following fMRI preprocessing, TaskFMRI extracts functional activation timecourses and maps from fMRI task data. Visual inspection and quality control will be performed using output from the Executive Summary feature of the pipeline to inspect individual surfaces, functional data registration, and motion related artifacts.

fMRI analysis.

A generalized linear model (GLM) within FSL will be run for each participant separately for each task. For the emotion regulation task, the three conditions (passive negative, passive neutral, and decreased negative) and all six motion parameters will be modeled with an implicit baseline. For the Stroop task, the three blocks (incongruent, congruent, and neutral) and all six motion parameters will be modeled with an implicit baseline. The parameter estimates from the GLM will be input into Permutation Analysis of Linear Models (PALM) to examine group differences in whole-brain activation while covarying for GA (Winkler et al., 2014). PALM uses a GLM to calculate a test statistic map. Z-score normalization is used on the resulting test statistic map and clusters are generated based on a threshold of Z = 2.3 to control for false positive rate (Eklund, Nichols, & Knutsson, 2016). The data are then permuted 10,000 times. Each permutation involves shuffling the data randomly with respect to the design matrix and repeating the analysis to extract the largest cluster size, resulting in a null distribution of cluster size across the 10,000 permutations. P values are calculated based on the proportional rank of each observed cluster size along the null distribution.

Power Analysis

Power analysis was conducted in PASS 13.0 (NCSS, 2017), taking into consideration anticipated attrition of 10% by the time of the six week postpartum visit and up to 10% missing data. For the proposed ANCOVA models, assuming one TAU group (N = 30), 5 MBCT-PD groups (N=6/group), 3 covariates, and α<0.05, we will have > 80% power to detect a mean difference of 0.8 when the three covariates have a combined R² of 0.35; power >80% to detect 1.0 SD difference when R² is close to 0. If applicable, mediation analyses using bias corrected bootstrap sampling will achieve power of .80 with N=60 given medium to larger effect sizes (Fritz & Mackinnon, 2007), as found in prior research (Dimidjian et al., 2015; Vieten & Astin, 2008).Prior fMRI studies of women with PPD have found group differences in frontal activation on an emotional responding task using significantly smaller group sizes (N = 11; Laurent & Ablow, 2013). Further, prior studies using the same fMRI Stroop task have found significant differences between clinical and non-clinical samples of 23 in each group (Banich et al., 2009).

Data Management

Data collection procedures.

All data collected on source forms will be double entered into REDCap. Examiner administered measures (e.g., SCID-5-CV) will be double-scored and then double entered. All data will be stored via a participant ID number. All source data forms will be stored separately from forms with identifying information. Video recordings will be stored on a secure server.

Assessment reliability.

All assessors will receive weekly supervision from the PI. Each month, 20% of recorded SCID-5-CV administrations will be randomly selected for inter-rater reliability rating. At least 85% diagnostic reliability must be maintained.

Intervention adherence.

Two recorded sessions will be selected at random from each MBCT-PD group and reviewed for fidelity by master’s level therapists using the MBCT Adherence Scale (MBCT-AS), a validated scale designed to assess fidelity to MBCT (Segal, Teasdale, Williams, & Gemar, 2002). Therapists must maintain a mean score > 1, which indicates adequate adherence.

Data Monitoring

There is a three-person independent monitoring committee (IMC) for the study. An independent monitor performed an initiation visit and will complete audits annually and at study completion; the monitor is independent from the investigators and NCCIH.

Open Science Plans

The trial is pre-registered on ClinicalTrials.gov, which was chosen to ensure compliance with NIH funding requirements. The study protocol, with documented amendments and versions, statistical analysis plan (SAP), and primary and secondary outcome results will be made available through ClinicalTrials.gov. The study design, protocol, and data analysis plan have been approved by the NCCIH. The NCCIH will not be involved in data management, data analysis, or drafting of study results. The PI will have access to the final trial dataset. Authorship will be granted to individuals making a significant scientific contribution to the drafted work; professional writers will not be used. Individual participant data (IPD) will be de-identified and shared with appropriate IRB and NCCIH approvals. Data will be available 12 months after final data collection. All peer-reviewed publications will be submitted to PubMed Central.