Brain Emotion Circuitry-Targeted Self-Monitoring and Regulation Therapy (BE-SMART): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Brain Emotion Circuitry-Targeted Self-Monitoring and Regulation Therapy (BE-SMART) (NCT03183388), a na na study. The primary outcome is Functional Magnetic Resonance Imaging (fMRI) Blood Oxygen-level Dependent (BOLD) Signal Changes in Left Amygdala at Baseline and at the End of the Intervention.

2. Clinical Question

The question is about a brain-imaging measure, not a direct measure of mood symptoms or daily functioning. The registry summary says that symptom, behavioral, and scanning assessments were planned before treatment, at midpoint, and after psychotherapy. It describes the intended use of this information as assessing emotion-regulation circuitry and functioning, but the primary outcome named in the record is the left amygdala BOLD signal.

Because there is no concurrent control group identified in the registry, a baseline-to-endpoint change cannot by itself distinguish an intervention-associated change from other explanations, such as changes over time or measurement variability. A comparison between the two named intervention approaches is also not described as randomized. The registry record does not report a prespecified statistical contrast between those approaches.

3. Trial Design

BE-SMART (NCT03183388) is listed as a completed, single-group behavioral study sponsored by Yale University. The registry reports an enrollment of 76. The study began on 2017-10-17, and primary completion was recorded as 2022-06-30. Phase is listed as NA, allocation as NA, and masking as NONE. These entries do not describe a blinded or randomized controlled trial.

The intervention is named BE-SMART. Its arm description characterizes it as a psychobehavioral intervention with a focus either on teaching emotional regulation skills or on regularizing daily sleep and activity levels. The primary outcome description refers to BE-SMART-DR and BE-SMART-ER separately. The registry record does not explain the assignment process for those approaches or report an external control group, so they should not be treated as randomized treatment and control arms.

The primary outcome timeframe is baseline and 12 weeks. The trial summary describes symptom and behavioral assessments and scanning before treatment, at midpoint, and after psychotherapy; it does not give the midpoint timing in the primary outcome timeframe. The registry lists four outcome measures and four analyses, but details about all of them are not reported in the registry record.

For the imaging outcome, participants completed an emotional face-processing task during fMRI. The record describes a voxel-level screening step followed by extraction of signal values from qualifying voxels in hypothesized regions of interest. It states that the only voxel-based finding meeting the specified criterion was a left amygdala activation decrease to fearful faces for BE-SMART-DR. The BE-SMART-ER voxel-level result did not reach the stated threshold, so the record says no values were extracted for that approach.

The registry does not report a sample-size justification, the intended allocation between the two intervention approaches, interim analyses, or a detailed statistical analysis plan. Any planning illustration based on an observed effect should be read as a hypothetical design calculation, not as the trial’s original design rationale or as observed power.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: 76 participants were enrolled and 60 started

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Adolescents and young adults with bipolar disorder (BD) who received intervention and had both baseline and endpoint fMRI data of sufficient quality were analyzed. Voxel-level threshold for BE-SMART-ER did not reach p\<0.001 uncorrected threshold and therefore there were no extracted values for further analyses.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Functional Magnetic Resonance Imaging (fMRI) Blood Oxygen-level Dependent (BOLD) Signal Changes in Left Amygdala at Baseline and at the End of the Intervention, measured over baseline and 12 weeks.

The primary endpoint is a continuous fMRI measure: BOLD signal in the left amygdala during an emotional face-processing task, evaluated at baseline and the end of the intervention. BOLD signal is an indirect imaging measure of changes in blood oxygenation associated with brain activity. In this study, the task used emotional faces, and the reported voxel-based finding concerned responses to fearful faces.

The endpoint is intended to quantify brain-circuit activity relevant to emotion processing. The study summary frames the broader objective as assessing whether the intervention could improve functioning of emotion-regulation brain circuitry. The recorded primary measure, however, is a signal difference in a specified brain region; it is not itself a direct measure of symptom improvement, functioning in daily life, or clinical benefit.

The analysis description says baseline-to-endpoint signal differences were screened voxel by voxel. Voxels passing the specified threshold within a hypothesized region of interest were extracted, and mean values within that region for each subject were used in the subsequent analyses. For the qualifying BE-SMART-DR finding, the registry describes a mixed-model analysis and identifies the reported parameter as a least-squares mean with standard error. The numerical estimate, standard error, confidence interval, and exact mixed-model test result are not reported in the registry record.

Using a regional imaging outcome makes it possible to focus on a hypothesized circuit rather than treating the entire scan as a single measurement. The interpretation nevertheless depends on the screening and extraction procedure, the imaging data quality, and the statistical model. The record does not provide enough detail to assess every part of that procedure.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry describes a multistep analysis of fMRI data collected during an emotional face-processing task. The outcome is continuous BOLD signal, measured in relation to baseline and the end of the intervention. The analysis description first applies a voxel-level threshold and then extracts signal values from qualifying voxels in hypothesized regions of interest, including the amygdala and ventral prefrontal cortex. It states that mean values within the region for each subject were used in subsequent analyses. For the left amygdala finding, the record describes a mixed-model analysis and identifies the result type as a least-squares mean with standard error.

What does a voxel-level threshold do?

General statistical explanation: an fMRI scan contains many spatially distinct locations, or voxels. Testing for a signal difference at each voxel creates a large family of statistical tests. A threshold is a rule for deciding which voxel-level results are carried forward for further consideration. The registry states a threshold of p < 0.001 uncorrected, using statistical parametric mapping software. The word “uncorrected” indicates that the threshold is not adjusted for the multiple voxel-level tests. That distinction is important because, when many tests are conducted, some may pass a threshold by chance even if there is no underlying signal at those locations.

The record reports that the only voxel-based finding meeting its criteria was a decrease in left amygdala activation to fearful faces for BE-SMART-DR. It also states that BE-SMART-ER did not reach the threshold and that no values were extracted for that approach. This is a result of the specified screening and extraction process; it is not evidence that every other brain region or intervention outcome had no change.

Why extract regional values?

General statistical explanation: a region-of-interest analysis can reduce a complex image to a smaller set of values with a defined anatomical focus. This can make the later analysis easier to interpret in relation to a proposed brain circuit. In this record, extraction followed the voxel-level screen, and the stated values were then summarized within the region for each subject. Because the extraction is conditional on passing the screen, the later regional analysis is connected to the initial voxel-level selection. The registry record does not report enough detail to evaluate the full implications of that selection process.

What is a least-squares mean?

General statistical explanation: a least-squares mean is a model-based mean, also called an adjusted mean in many settings. Its value depends on the statistical model and the terms included in that model. The registry names a least-squares mean but does not report the model specification, covariates, covariance structure, or the numerical estimate. It is therefore not possible to reconstruct what adjustment was made or to assess the estimate’s magnitude from the registry description alone.

How should the standard error be read?

General statistical explanation: a standard error describes the estimated sampling uncertainty of a model-based estimate. It is not the same as the spread of individual participant measurements. A smaller standard error generally indicates greater precision for the estimated mean, while a larger standard error indicates less precision. In this record, the parameter and dispersion are identified as least-squares mean and standard error, but the actual values are not reported in the registry record. A confidence interval, which would show a range of estimates compatible with the data under the model, is also not reported.

Why might a mixed model be used?

General statistical explanation: mixed models can account for data that are related within participants or organized into groups, depending on how the model is specified. They can also represent repeated measurements when data are available at more than one time. The registry says a mixed model was used for the extracted values, but it does not state which fixed or random effects were included, how repeated measurements were handled, or which assumptions were evaluated. Those details should not be inferred from the label alone.

What does over-dispersion mean in general?

General statistical explanation: over-dispersion occurs when observed variability is greater than a model’s assumed variability. If it is present and not addressed, uncertainty estimates can be too small, making results appear more precise than they are. This is a general issue for some statistical models; the registry does not report an over-dispersion assessment, and the continuous fMRI outcome description does not establish that over-dispersion was relevant to this analysis.

What do planning calculations represent?

General statistical explanation: sample-size planning calculations describe how many participants might be needed to detect a chosen effect under specified assumptions. The registry’s planning illustration is explicitly based on the observed effect, with a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and effects defined as stated fractions of the observed effect. These are planning assumptions, not a description of the actual single-group design. Such calculations are not observed power and do not show that the study had a particular probability of detecting an effect after data collection.

10. Multiplicity, Missing Data and Analysis Populations

The registry describes a voxel-level screening threshold of p < 0.001 uncorrected. It does not report a multiple-testing correction for the voxelwise analyses, nor does it state how multiplicity across the listed outcome measures or analyses was handled. In general, testing many locations or outcomes increases the chance of obtaining at least one apparently notable result by chance. Correction procedures and prespecified analysis priorities can help address that issue, but their use should not be assumed when the record does not report them.

The record gives a primary outcome description and says four outcome measures and four analyses were listed. The complete outcome and analysis details are not reported in the registry record. It is also not reported whether the reported imaging analysis was prespecified in a detailed analysis plan, whether any analyses were exploratory, or whether interim analyses were conducted.

The imaging population is described as participants with baseline and endpoint fMRI data of sufficient quality. This indicates that the imaging analysis did not necessarily include everyone enrolled, but the number analyzed for this endpoint is not reported in the registry record. The reason participants might lack usable data, and whether any missing-data methods were used, are not reported in the registry record. The registry summary mentions assessments at baseline, midpoint, and after psychotherapy, but does not describe how incomplete symptom, behavioral, or imaging observations were handled.

In general, missing data can affect both the composition of the analyzed group and the precision of estimates. Complete-case analyses may differ from analyses that use all available observations, and the consequences depend on why data are missing. Since the registry does not report the missing-data strategy or detailed analysis populations, the impact of missing or unusable measurements cannot be assessed from the record.

11. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

The registry reports 60 participants at risk for the BE-SMART safety group. It records 0 serious adverse events, 0 other adverse events, and 0 deaths in that group. No top serious adverse event terms are listed. These are the posted safety counts for the group and should be read as registry-reported event figures, not as a guarantee that no events of any kind occurred outside the collection and reporting represented by the record.

The enrollment figure is 76, while the safety group’s at-risk count is 60. These figures describe different registry entries and should not be treated as interchangeable denominators. The record does not explain the difference between the enrolled group and the safety group at risk, so its cause cannot be inferred. The fMRI analysis population is separately described as participants with sufficiently good baseline and endpoint scans, but its size is not reported in the registry record.

Zero recorded events is a count, not a precise estimate of the underlying risk. In general, rare events can remain compatible with a dataset in which no events were observed, particularly when the monitored group is limited. The registry does not report additional details about safety surveillance, event definitions, collection procedures, or follow-up beyond the at-risk and affected counts. The posted figures should therefore be interpreted within those reporting limits.

12. What the Result Does, and Does Not, Mean

The reported result is a voxel-based left amygdala activation decrease to fearful faces among participants receiving BE-SMART-DR, meeting the stated p < 0.001 uncorrected threshold. The registry says this was the only voxel-based finding meeting criteria. It also reports that the BE-SMART-ER result did not reach the threshold and yielded no extracted values for subsequent analysis. These statements describe the outcome of the reported screening and analysis procedure.

The record does not provide the estimated size of the left amygdala change, its standard error, a confidence interval, or an exact model-based p-value. The numerical mixed-model results are not reported in the registry record. Consequently, the magnitude and precision of the extracted regional change cannot be judged from this record, and the threshold result should not be mistaken for a complete account of the effect estimate.

The study design limits causal interpretation. It is a single-group study with no concurrent control group identified in the registry. A change between baseline and endpoint may be associated with participation in the study, but the design does not isolate the intervention’s effect from changes that could occur over time or through other influences. The record also does not report randomized allocation between BE-SMART-DR and BE-SMART-ER, so their results do not constitute a controlled comparison.

The endpoint is a brain-imaging signal rather than a direct clinical outcome. The study summary describes symptom and behavioral assessments, but their numerical results are not reported in the registry record. The imaging finding therefore cannot establish that symptoms, behavior, or everyday functioning changed. Nor can the absence of a qualifying BE-SMART-ER voxel result establish that there was no underlying change.

In general, an imaging result can be informative about a hypothesized brain measure while leaving clinical importance, reproducibility, and causal attribution unresolved. Those questions require interpretable effect estimates, appropriate uncertainty measures, complete reporting of analysis choices, and a design capable of addressing the relevant comparison. The registry record does not contain enough information to resolve them.

13. Why This Trial Matters Statistically

This study illustrates how a clinical intervention can be evaluated using a mechanistic imaging outcome. BE-SMART was intended to teach emotional regulation skills or regularize daily sleep and activity levels, while the primary outcome focused on left amygdala BOLD signal during emotional face processing. That distinction matters statistically: a measure can address a proposed brain process without directly measuring symptoms or clinical functioning.

The analysis also demonstrates how imaging workflows can depend on sequential decisions. The registry describes voxelwise screening followed by regional signal extraction and a mixed-model analysis. A result at the screening stage determines which regional values are taken forward. Readers need information about thresholding, multiplicity, model specification, and uncertainty to judge how robustly that sequence supports the reported finding. The record provides the uncorrected threshold and analysis labels but not the detailed model or numerical estimates.

Design and reporting are equally important. A single-group baseline-to-endpoint analysis can describe change, but it does not provide the same causal comparison as a randomized controlled design. The registry identifies no like-for-like control group and does not report random allocation between the intervention approaches. This makes it especially important to avoid interpreting a within-group change as an intervention effect.

The record also highlights the value of transparent reporting of analysis populations and missing data. The fMRI analysis required usable scans at both baseline and endpoint, yet the analyzed sample size and handling of incomplete data are not reported. Without those details, readers cannot assess how the imaging analysis population relates to overall enrollment or how selection into analysis may affect interpretation.

14. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The primary limitation is that the registry record does not report the numerical mixed-model estimate for the primary imaging outcome. The least-squares mean, standard error, confidence interval, and exact model-based test result are not reported in the registry record. The direction of the voxel-based finding is stated, but its size and precision cannot be evaluated.

The imaging result is based on voxel-level screening at p < 0.001 uncorrected. The registry does not report a correction for the many voxel-level tests or how multiplicity across outcome measures and analyses was addressed. It also does not report the detailed statistical analysis plan, the mixed-model specification, or whether interim analyses were conducted.

The single-group design has no concurrent control group identified in the registry. It therefore cannot distinguish the observed baseline-to-endpoint change from other changes over time. Assignment to the two described BE-SMART approaches is not reported, and the record does not establish a randomized comparison between them.

The imaging analysis required baseline and endpoint fMRI data of sufficient quality, but the analyzed sample size is not reported. The registry does not describe the amount or pattern of missing or unusable data, the handling of missing observations, or whether the analyzed participants differed from those not included in the fMRI analysis. The full population description in the outcome entry is not reported in the registry record.

Finally, the reported endpoint is an imaging measure. Although the study summary refers to symptom and behavioral assessments, their results are not reported in the registry record. The safety record lists 60 participants at risk, but the reason this differs from the enrollment of 76 is not reported. These omissions restrict assessment of the findings’ precision, generalizability, and relationship to clinical outcomes.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Brain Emotion Circuitry-Targeted Self-Monitoring and Regulation Therapy (BE-SMART) — NCT03183388.
• Registry record last updated 2024-02-29; status last verified 2024-02.
• Results first posted 2024-02-29.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

18. Record Summary

NCT03183388 is a completed, single-group behavioral study of BE-SMART sponsored by Yale University, with enrollment of 76. The study began on 2017-10-17, and primary completion was recorded as 2022-06-30. The intervention focused either on emotional regulation skills or on regularizing daily sleep and activity levels.

The primary outcome was left amygdala BOLD signal during an emotional face-processing task, assessed at baseline and 12 weeks. The registry reports a voxel-based decrease in activation to fearful faces for BE-SMART-DR as the only finding meeting p < 0.001 uncorrected. BE-SMART-ER did not reach that threshold. The numeric mixed-model estimate and its uncertainty are not reported in the registry record.

The safety group lists 60 participants at risk, 0 serious adverse events, 0 other adverse events, and 0 deaths. No concurrent control group, detailed model specification, analysis sample size, or missing-data strategy is reported in the registry record.