Effects of Vitamin D and Omega-3 on Cerebrovascular Disease: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Effects of Vitamin D and Omega-3 on Cerebrovascular Disease (NCT04070833), a randomized na study. The primary outcome is Number of Participants With Poor Stroke Outcome at Hospital Discharge as Measured by the Modified Rankin Scale (mR 3-6).

2. Clinical Question

The study concerns outcomes after stroke among participants in the parent VITAL study, rather than the risk of experiencing a first stroke in the general population. The registry description says that VITAL participants who experienced a stroke during follow-up are the population for this study, which examines whether the randomized supplement assignments are associated with post-stroke outcomes.

The outcome is a binary classification of the modified Rankin Scale (mRS): scores from 0 to 2 are described as good, and scores from 3 to 6 as poor. The primary endpoint specifically counts participants in the poor category at hospital discharge. Although the record defines the question and lists four treatment combinations, it does not include the counts of poor outcomes by group or a reported comparison. Consequently, the clinical question is clear, but the posted record does not support a conclusion about whether the groups differed.

3. Trial Design

The registry classifies the study as randomized, triple-masked, and factorial. Enrollment is recorded as 290. The sponsor is Brigham and Women's Hospital. The record lists the study start as 2016-08-01, primary completion as 2018-09-17, and status as completed. The phase is listed as not applicable.

The four groups combine assignments for vitamin D and fish oil: vitamin D plus fish oil, vitamin D plus fish oil placebo, vitamin D placebo plus fish oil, and both placebos. This arrangement is consistent with the stated factorial model. In general, a factorial design can examine the effects of more than one intervention factor within a single study. Whether the analyses estimated each factor's effect, evaluated an interaction, or used a different approach is not reported in the registry record.

The registry describes this study as involving participants in the completed VITAL trial who experienced a stroke event during follow-up. It notes that the parent study examined daily vitamin D3 and omega-3 supplementation in people without a prior history of the illnesses named in its description, and that observational follow-up was ongoing. Those details describe the broader parent-trial context; they do not provide a statistical analysis plan for this stroke-outcome study.

The primary endpoint is poor mRS outcome at hospital discharge, assessed at a timeframe recorded as 1 day. The record describes the mRS as a seven-point scale, from 0, no symptoms, through 6, death, and explains that the analysis would categorize scores into good and poor outcomes to avoid convergence problems from sparse data. The actual group-specific endpoint counts, analysis population size, and treatment-effect estimates are not reported. The registry record also lists five outcome measures and zero analyses.

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: This study only enrolled individuals from the main VITAL study who experienced a stroke during the main VITAL study. Therefore, although the main VITAL study enrolled 25871 individuals, the current study only includes the 290 individuals who experienced a stroke event during the main VITAL study.

Overall Study

Analysis population for the primary outcome, as defined in the registry: The overall Number of Participants Analyzed is less than the number of participants in the Participant Flow module because the mRS was missing for some participants.

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 Number of Participants With Poor Stroke Outcome at Hospital Discharge as Measured by the Modified Rankin Scale (mR 3-6), measured over 1 day.

The primary endpoint counts participants with poor stroke outcome at hospital discharge, defined as an mRS score from 3 to 6. The registry describes the mRS as a measure of disability or a global health index, with strong emphasis on mobility. Its score ranges from 0, no symptoms, to 6, death; scores 0 to 2 are categorized as good outcomes and scores 3 to 6 as poor outcomes for this endpoint.

Reducing the scale to two categories can make the target outcome easier to describe as a participant count. The registry specifically says the categorization was planned to avoid model-convergence issues associated with sparse data. In general, convergence refers to whether an estimation procedure can find a stable solution under the specified model. Sparse data can make that harder, particularly when a model attempts to estimate many categories or parameters from limited observations.

The endpoint is assessed at hospital discharge, with a recorded timeframe of 1 day. Its stated unit is participants, and the measure is a count rather than a continuous score or time-to-event measure. The registry notes that the mRS was collected during the parent study and that some participants had missing mRS data, so the number analyzed was less than the number in participant flow.

The registry record defines the endpoint but does not report the number of participants in either outcome category, counts by randomized group, or an effect estimate. Thus, the endpoint definition explains what would be counted, but the posted record does not show the observed primary-outcome comparison.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry labels the allocation randomized, the masking triple, and the model factorial. Randomization is a method for assigning participants to study groups by chance. In general, it is intended to make groups comparable on average at assignment, although it does not guarantee that every measured or unmeasured characteristic will be balanced. Masking is intended to limit the influence of knowledge about assignments on study conduct or assessment. The record does not specify which roles were masked, how masking was maintained, or how allocation was implemented.

A factorial design studies multiple intervention factors in the same framework. The four listed groups combine vitamin D assignment with fish oil assignment, including placebo assignments for each. In general, such a design can estimate effects associated with each factor and can assess whether the effect of one factor depends on the other. That dependence is commonly called an interaction. The registry's use of the word factorial does not, by itself, reveal which effects were estimated or how the interaction was handled. Those details are not reported in the registry record.

The primary measure is a participant count for poor outcome at hospital discharge. The mRS is described as a seven-point disability scale, with the registry recording values from 0 through 6. The primary endpoint combines scores 3 to 6 as poor and scores 0 to 2 as good. In general, dichotomizing an ordinal scale makes the target easier to communicate as a binary outcome, but it also removes distinctions among scores within each category. For example, the endpoint treats all scores in the poor category as the same classification for the primary count, even though the scale's levels represent different states.

The registry says this categorization was planned to avoid convergence problems due to sparse data. Generally, a model may have difficulty producing stable estimates when outcome categories contain few observations relative to the model's parameters. Combining categories can reduce the number of outcome levels that a model must handle. This can improve computational stability, but the choice of cut point also determines which outcomes are grouped together. The record explains the rationale but does not report the model ultimately used or demonstrate whether convergence was achieved.

For a binary participant-count outcome, one possible broad analytic approach is to compare the proportions of participants with the event across groups. Other approaches can adjust for covariates or model multiple factors jointly. These are general possibilities, not descriptions of this study's reported analysis. The registry lists zero analyses and does not identify a statistical test, model, covariate set, estimand, or effect measure. No particular method should be attributed to the study based on the design label alone.

In general, an effect estimate is a numerical summary of a comparison, and its interpretation depends on the scale. A difference in proportions, a ratio of proportions, and an odds-based measure answer related but not identical questions. A confidence interval describes uncertainty around an estimate under the assumptions of the analysis; it is not a guarantee that a specific range contains the true effect. Neither the primary effect estimate nor its confidence interval is reported in this record.

The registry includes sample-size assumptions for planning illustrations based on the observed effect, including a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. Such illustrations are planning calculations rather than observed trial findings. The record does not report the trial's own sample-size justification, whether those assumptions formed part of the original design, or an observed-power analysis. No planning illustration can replace a missing outcome estimate or analysis.

The listed outcome timeframe is 1 day, while the endpoint description specifies mRS at hospital discharge. The record indicates that the mRS data were collected during the parent study. It does not describe additional assessment timing, adjudication procedures, or outcome-assessor training. The recorded timeframe and outcome definition can identify the intended assessment, but they do not provide the assessment counts or explain the operational handling of all cases.

Five outcome measures are listed in the registry, but the posted record does not give a full analysis plan for them. In general, a study with multiple outcomes or contrasts must define which analyses are primary and how interpretation accounts for multiple opportunities to find a result. The record identifies a primary endpoint, yet it does not describe a multiplicity strategy, a hierarchy among outcomes, or any prespecified inferential procedure. The statistical plan is not reported in the registry record.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists five outcome measures and identifies poor mRS outcome at hospital discharge as the primary endpoint. It does not report a statistical analysis plan, an inferential procedure, or a strategy for handling multiplicity. In general, multiplicity arises when several outcomes, contrasts, or analyses are considered; interpreting each without accounting for the broader set can make isolated findings harder to evaluate. The record does not say whether or how such issues were addressed.

The record also indicates that some mRS values were missing. It states that the overall number analyzed was less than the number in participant flow because the mRS was missing for some participants. The number with missing values, their distribution across treatment groups, and the method used to handle missing outcomes are not reported in the registry record. No assumption about the reasons for missingness or its likely effect on the comparison can be made from the entry.

The record gives enrollment as 290, but it does not provide a primary-endpoint analysis population count or a participant flow specific to the endpoint. The enrollment figure should not be treated as the number with an observed mRS score. The record separately identifies four randomized treatment combinations but does not report group-specific endpoint denominators.

General approaches to missing data can rely on different assumptions and may produce different estimates. Without the number and pattern of missing scores, and without the analysis method, readers cannot assess whether the endpoint comparison could be sensitive to missingness. Similarly, without a stated multiplicity procedure, the handling of the several listed outcomes cannot be evaluated. These are reporting gaps; they do not establish that a particular analysis was or was not conducted outside the posted registry 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.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: major cvd (30.5%), malignant cancer (3.7%), gastrointestinal bleeding (3.0%), hypercalcemia (1.5%).

The registry lists serious adverse events, other adverse events, and deaths for each of the four treatment combinations. For serious events, it records the number at risk and the number affected in each group; it does the same for other adverse events. The posted values are counts, not adjusted comparisons. They should be read as descriptive safety information rather than as proof that one assignment caused more or fewer events.

For serious adverse events, the record lists 66 at risk and 18 affected in the vitamin D plus fish oil group; 75 at risk and 20 affected in the vitamin D plus fish oil placebo group; 82 at risk and 28 affected in the vitamin D placebo plus fish oil group; and 67 at risk and 19 affected in the vitamin D placebo plus fish oil placebo group. Death counts are listed as 7, 4, 7, and 4, respectively, in the same group order. The record does not report a statistical comparison of these counts.

It also names major CVD at 30.5%, malignant cancer at 3.7%, gastrointestinal bleeding at 3.0%, and hypercalcemia at 1.5% among the top serious-event terms. The record does not explain in this entry how these terms relate to each participant's event history or whether categories can overlap. They should not be added or interpreted as mutually exclusive categories. The serious-event figures are distinct from the primary mRS endpoint, and the registry provides no causal or adjusted safety analysis.

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

The record describes a completed randomized, triple-masked factorial study of vitamin D and fish oil among participants in VITAL who experienced a stroke during follow-up. It specifies the primary outcome as the number of participants with mRS scores from 3 to 6 at hospital discharge. It also reports safety counts for the four treatment combinations. These details establish the study's context and intended outcome, but they do not provide a primary treatment-effect result.

No group-specific primary endpoint counts, statistical estimates, confidence intervals, or p-values are reported. Therefore, the registry record does not show how the groups compared on poor stroke outcome, how precise any comparison was, or whether a formal analysis found evidence of a difference. The absence of a posted estimate is not evidence that the groups were equivalent; it means the result cannot be judged from the record's reported primary-outcome information.

The endpoint's binary classification is also narrower than the full mRS scale. It combines scores from 3 through 6 as poor and scores from 0 through 2 as good. As a general statistical matter, this preserves the distinction across the specified cut point but does not retain all the scale's gradations in the endpoint count. The registry explains that the categorization was intended to address sparse-data convergence concerns.

The serious adverse-event and death counts give descriptive information by treatment group, but the record does not report inferential comparisons or adjusted safety estimates. Those counts cannot be used to infer causation or to answer the primary endpoint question. The registry also notes missing mRS values, without giving their number or the method used to handle them.

Accordingly, the defensible conclusion is limited to what is documented: the study design, endpoint definition, and listed adverse-event counts. The registry does not provide enough primary-outcome analysis information to determine the size, direction, or precision of an effect on poor stroke outcome.

13. Why This Trial Matters Statistically

This study illustrates how a registry entry can define a clinically interpretable endpoint while leaving the core statistical comparison unavailable. The mRS categorization turns a seven-point disability scale into a participant count for poor outcome at hospital discharge. That makes the intended endpoint explicit, and the stated concern about sparse data explains why the registry planned a binary grouping. But an endpoint definition cannot answer whether randomized groups differed; outcome counts and an analysis are also needed.

The factorial arrangement is statistically relevant because the four groups combine assignments for two supplement factors. In general, factorial studies can address more than one intervention contrast and can examine whether factors interact. The record's design label therefore raises important interpretation questions: which effects were estimated, what comparison each estimate represents, and how any interaction was treated. Those questions cannot be resolved from the registry information because the analyses are not reported.

Missing mRS data also matter to interpretation. The record expressly says fewer participants were analyzed than appeared in participant flow, but does not report the number missing or how missing outcomes were handled. Statistical conclusions can depend on these details, especially if missing data are unevenly distributed or if analysis assumptions are not met. Without this information, readers cannot assess the robustness of a primary endpoint comparison.

The record does contain adverse-event counts and names several serious-event terms. These illustrate why safety reporting and efficacy or outcome analysis need to be kept conceptually separate. Descriptive counts provide context, but without a formal comparison they do not establish differences attributable to treatment. Overall, the study matters statistically as a case where design and endpoint details are documented more fully than the inferential results needed to quantify the central question.

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 of the registry record for interpreting the main question is that it does not report primary endpoint results by group. Although the outcome definition and study design are recorded, the poor-outcome counts, denominators, effect estimate, confidence interval, and test result are not reported in the registry record. The record lists zero analyses, so the statistical evidence for the stated primary comparison cannot be assessed from this entry.

The endpoint analysis population is also unclear. The record notes that some mRS values were missing and that the number analyzed was below the number in participant flow, but it does not state how many values were missing, how many participants contributed to each group, or how missing scores were handled. This prevents evaluation of the extent and potential consequences of missing outcome data.

Although the study is identified as factorial, the actual analysis of the vitamin D and fish oil assignments is not described. It is not reported whether the primary endpoint was analyzed as a factorial main effect, through comparisons of the four combinations, or using another specified model. The handling of possible interaction, covariate adjustment, and multiplicity is also not reported.

The registry lists five outcome measures but does not describe their analysis hierarchy or a multiplicity approach. Nor does it report an analysis plan, sample-size justification, or interim analyses. These omissions do not show that such procedures were absent; they mean that they are not reported in the registry record. The listed adverse-event counts are descriptive, with no statistical comparison or adjusted safety analysis reported. Finally, the record's parent-trial context does not substitute for results specific to post-stroke outcomes in this study.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Effects of Vitamin D and Omega-3 on Cerebrovascular Disease — NCT04070833.
• Registry record last updated 2022-08-15; status last verified 2022-08.
• Results first posted 2022-08-15.
• 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

NCT04070833 is a completed, randomized, triple-masked factorial study sponsored by Brigham and Women's Hospital, with enrollment recorded as 290. The study concerns VITAL participants who experienced a stroke during follow-up and combines vitamin D and fish oil assignments with their placebos across four groups.

The primary endpoint is the number of participants with poor mRS outcome at hospital discharge, defined as scores from 3 to 6. The registry says mRS values were missing for some participants, making the number analyzed lower than the number in participant flow. It lists five outcome measures and zero analyses.

Serious adverse-event counts, other adverse-event counts, deaths, and selected serious-event terms are recorded. However, primary endpoint counts by group, treatment-effect estimates, uncertainty intervals, and inferential comparisons are not reported in the registry record. The posted entry therefore defines the clinical question and records descriptive safety information, but does not allow interpretation of a primary treatment-effect estimate.