Effects of cocoa extract and a multivitamin on cognitive function: A randomized clinical trial

Table of Contents

Overall Summary

Study Background and Main Findings

The COSMOS-Mind study, a large, randomized, two-by-two factorial, 3-year trial, found that daily multivitamin-mineral (MVM) supplementation significantly improved global cognition in older adults (mean z-score change = 0.07, 95% CI: 0.02 to 0.12; P = .007) compared to placebo. This positive effect was particularly pronounced in participants with a history of cardiovascular disease (P = .01). However, daily cocoa extract supplementation had no significant effect on global cognition (mean z-score change = 0.03, 95% CI: -0.02 to 0.08; P = .28).

Research Impact and Future Directions

The COSMOS-Mind study provides compelling evidence that daily MVM supplementation can improve global cognition in older adults, particularly those with a history of cardiovascular disease. The study's rigorous design, large sample size, and comprehensive cognitive assessment are significant strengths. However, the study also makes clear distinctions between correlation and causation. While MVM supplementation was associated with improved cognition, the study design does not allow for definitive causal claims, as unmeasured confounding factors may have influenced the results.

The practical utility of the findings is substantial, suggesting that MVM supplementation may be a readily accessible and low-cost intervention to support cognitive health in older adults. These findings are placed within the context of existing research, which has shown mixed results regarding the effects of individual micronutrients on cognition. The current study adds to this body of knowledge by demonstrating the potential benefits of a comprehensive MVM approach.

While the findings are promising, several uncertainties remain. The study's reliance on a predominantly non-Hispanic White population limits the generalizability of the results to other demographic groups. Additionally, the precise mechanisms underlying the observed cognitive benefits are not fully elucidated. Future research should focus on replicating these findings in more diverse populations and exploring the role of specific micronutrients and biological pathways.

Critical unanswered questions include the optimal dosage and duration of MVM supplementation for cognitive benefits, the long-term effects of supplementation, and the potential interactions between MVMs and other lifestyle factors. While the methodological limitations, such as the lack of diversity and reliance on self-reported data for some measures, do not fundamentally undermine the main conclusions, they do highlight the need for further research to confirm and extend these findings. Future studies should also address these limitations to provide a more comprehensive understanding of the role of MVM supplementation in promoting cognitive health in older adults.

Critical Analysis and Recommendations

Significant Improvement in Global Cognition with MVM (written-content)
Daily MVM supplementation resulted in statistically significant improvements in global cognition compared to placebo. This finding is important because it provides the first evidence from a large, long-term trial that MVMs may be a readily accessible intervention to improve cognitive function in older adults.
Section: Abstract
Enhanced MVM Benefit in Cardiovascular Disease Subgroup (written-content)
The positive effect of MVM supplementation was more pronounced in participants with a history of cardiovascular disease. This matters because it identifies a specific population that may derive greater cognitive benefits from MVM supplementation, potentially due to addressing underlying vascular-related cognitive impairments or micronutrient deficiencies.
Section: Highlights
Rigorous Study Design (written-content)
The study employed a randomized, two-by-two factorial design, which allowed for the simultaneous assessment of two interventions (cocoa extract and MVM). This is a major strength because it enhances the study's efficiency and internal validity by minimizing bias and allowing for the investigation of potential interactions between interventions.
Section: 2 METHODS
Comprehensive Cognitive Assessment (written-content)
The study utilized a comprehensive battery of cognitive tests, administered annually over three years. This is important because it allowed for a detailed evaluation of different cognitive domains, including global cognition, episodic memory, and executive function, providing a more complete picture of the interventions' effects.
Section: 2 METHODS
Lack of Detail on MVM Composition in Abstract (written-content)
The Abstract does not specify the MVM composition. This lack of detail hinders the reader's ability to understand precisely what intervention was tested and limits the comparability of the findings with other studies that may have used different MVM formulations.
Section: Abstract
Limited Discussion of Mechanisms in Discussion (written-content)
The Discussion section provides a limited exploration of the potential mechanisms underlying the observed effects of MVM supplementation. Expanding this discussion to include pathways such as oxidative stress reduction, inflammation, and vascular health, and exploring the potential role of specific micronutrients, would provide a more comprehensive theoretical framework and generate more specific hypotheses for future research.
Section: 4 DISCUSSION
Lack of Diversity in Study Population (written-content)
The study population was predominantly non-Hispanic White (89%). This is a limitation because it restricts the generalizability of the findings to other racial and ethnic groups, who may have different nutritional needs and responses to MVM supplementation.
Section: 4 DISCUSSION
Absence of P-values in Figure 2 (graphical-figure)
Figure 2 presents the three-year change in global cognition composite by treatment assignment but lacks p-values or tests for statistical significance. Including p-values would allow for a more definitive assessment of the statistical significance of the observed differences between groups, particularly in the subgroup analyses.
Section: 3 RESULTS

Section Analysis

Abstract

Key Aspects

Strengths

Suggestions for Improvement

Highlights

Key Aspects

Strengths

Suggestions for Improvement

1 BACKGROUND

Key Aspects

Strengths

Suggestions for Improvement

2 METHODS

Key Aspects

Strengths

Suggestions for Improvement

3 RESULTS

Key Aspects

Strengths

Suggestions for Improvement

Non-Text Elements

FIGURE 1 Consort diagram showing flow of participants from first approach...
Full Caption

FIGURE 1 Consort diagram showing flow of participants from first approach through randomization to each of the four treatment combinations in the two-by-two factorial design. Abbreviations: CE, cocoa extract; MVM, multivitamin-mineral

Figure/Table Image (Page 5)
FIGURE 1 Consort diagram showing flow of participants from first approach through randomization to each of the four treatment combinations in the two-by-two factorial design. Abbreviations: CE, cocoa extract; MVM, multivitamin-mineral
First Reference in Text
Figure 2A shows change in global cognition, relative to baseline, for participants assigned to CE (Figure 1, Groups 2 & 4) versus CE-placebo (Figure 1, Groups 1 & 3).
Description
  • Flow of Participants: The diagram is a flow chart that illustrates the journey of participants in the study, starting from when they were first approached to take part in the research. It shows how many people were initially contacted, how many were excluded at different stages, and for what reasons. For example, some people might not have been contactable, some might have declined to participate, and others might have been excluded for not meeting specific criteria.
  • Randomization: The diagram outlines the process of 'randomization,' which is a method used to fairly assign participants to different treatment groups. This is like drawing names out of a hat to decide who gets which treatment, ensuring no bias in how people are assigned. In this study, randomization ensures that each participant has an equal chance of being assigned to any of the treatment groups, which is crucial for the scientific validity of the results because it helps to ensure that any differences observed between the groups are due to the treatment itself, and not some other factor.
  • Two-by-Two Factorial Design: The study uses a 'two-by-two factorial design.' This is a type of experimental setup where two different treatments are tested both separately and together. It's like testing whether a plant grows better with fertilizer A, fertilizer B, both, or neither. Here, the treatments are cocoa extract (CE) and a multivitamin-mineral (MVM) supplement. This design allows researchers to see not only if each treatment works on its own but also if they have a combined effect when given together.
  • Four Treatment Combinations: The diagram shows that participants are divided into four groups, each receiving a different combination of treatments: (1) CE-Placebo and MVM-Placebo, meaning they get neither the cocoa extract nor the multivitamin; (2) CE and MVM-Placebo, meaning they get the cocoa extract but not the multivitamin; (3) CE-Placebo and MVM, meaning they get the multivitamin but not the cocoa extract; and (4) CE and MVM, meaning they get both treatments. 'Placebo' is a dummy treatment that has no active effect, used as a control to compare against the actual treatments.
  • Completion of Assessments: The chart indicates how many participants completed each stage of the study, specifically the baseline cognitive assessment and subsequent annual assessments. 'Cognitive assessment' refers to tests that measure mental functions like memory and thinking skills. The numbers show how many people dropped out or were unable to complete the assessments each year. For example, it shows that out of 587 participants in the CE-Placebo, MVM-Placebo group, 462 completed the Year 3 assessment, while 125 did not.
Scientific Validity
  • Comprehensive Participant Tracking: The CONSORT diagram meticulously tracks participant flow, which is crucial for assessing the study's internal validity. It provides a clear account of attrition rates and reasons for withdrawal, which are essential for evaluating potential biases. However, the reasons for non-contactability and refusal are not detailed, which could be important for understanding selection bias.
  • Randomization Integrity: The diagram confirms that participants were randomized, which is a strength of the study design. However, it does not detail the method of randomization or allocation concealment, which are important for ensuring the integrity of the randomization process.
  • Factorial Design Appropriateness: The use of a two-by-two factorial design is appropriate for the research question, allowing for the assessment of both main effects and interactions between CE and MVM. This design is well-suited to investigate the independent and combined effects of the interventions.
Communication
  • Clarity of Group Labeling: The diagram clearly labels each of the four treatment groups, aiding in the understanding of the study's design. The use of abbreviations is defined in the caption, which is helpful.
  • Visual Presentation of Attrition: The diagram effectively uses a visual format to present the flow of participants, making it easy to follow the progression and attrition at each stage. However, the lack of specific reasons for withdrawal or non-completion at each stage limits the depth of information conveyed.
  • Accessibility of Information: While the diagram is informative, it could be improved by providing more context within the figure itself, such as brief explanations of the exclusion criteria or reasons for dropout. This would enhance the stand-alone interpretability of the figure without needing to refer to the main text.
FIGURE 2 Three-year change in global cognition composite by assignment to (A)...
Full Caption

FIGURE 2 Three-year change in global cognition composite by assignment to (A) daily cocoa extract supplementation, and (B) daily multivitamin-mineral supplementation (plotted values: mean standardized (z) scores (relative to baseline) and 95% confidence intervals). Forest plot marginal differences by baseline subgroups for (C) cocoa extract and (D) multivitamin-mineral treatment groups

Figure/Table Image (Page 8)
FIGURE 2 Three-year change in global cognition composite by assignment to (A) daily cocoa extract supplementation, and (B) daily multivitamin-mineral supplementation (plotted values: mean standardized (z) scores (relative to baseline) and 95% confidence intervals). Forest plot marginal differences by baseline subgroups for (C) cocoa extract and (D) multivitamin-mineral treatment groups
First Reference in Text
As seen in Figure 3, at base- line, participants without CVD history outperformed those with CVD history on the global cognition composite (mean difference = 0.22, 95% CI: 0.08 to 0.37; P = .003); after Year 1, the MVM-placebo declined while the MVM group showed relative improvement (or pro- tection against decline).
Description
  • Change in Global Cognition Over Three Years: The figure illustrates how participants' thinking and memory skills, referred to as 'global cognition,' changed over a three-year period. This change is measured using a 'global cognition composite,' which is a combined score from various tests that assess different aspects of brain function, like memory, attention, and problem-solving abilities. The results are shown separately for two different treatments: daily cocoa extract supplementation and daily multivitamin-mineral supplementation.
  • Standardized (z) Scores: The changes in cognition are presented as 'standardized (z) scores.' A z-score is a way of representing data that shows how far a particular data point is from the average (mean), measured in terms of standard deviations. Think of it like a common yardstick that allows for comparison across different tests or measurements. Here, it indicates how much each participant's cognitive performance changed from their own starting point (baseline).
  • 95% Confidence Intervals: The '95% confidence intervals' provide a range within which we can be 95% certain the true average change in cognition lies. It's like saying, 'We're pretty sure the real answer is between this number and that number.' In a graph, these intervals are often shown as bars extending from the plotted points, indicating the precision of the estimate.
  • Forest Plots for Subgroup Differences: The figure includes 'forest plots,' which are a way of graphically representing the results of different studies or subgroups within a study. In this case, they show how the effects of cocoa extract and multivitamin-mineral supplementation differ across various subgroups of participants. These subgroups are defined by characteristics like age, sex, body mass index, and medical history. A forest plot typically displays the effect size (here, the difference in cognitive change) for each subgroup, along with its confidence interval, allowing for a visual comparison of treatment effects across different groups.
  • Marginal Differences: The term 'marginal differences' refers to the average difference in the outcome (change in global cognition) between the treatment groups (active vs. placebo) within each subgroup. It's a way of looking at the treatment effect separately for each group, like comparing the effect of a medicine on men versus women or younger versus older adults. This helps to see if the treatment works differently for different types of people.
Scientific Validity
  • Appropriateness of Outcome Measure: The use of a global cognition composite is appropriate for assessing overall cognitive change. However, the specific tests contributing to this composite should be scrutinized for their relevance and sensitivity to the interventions being studied. The reliance on z-scores is standard practice, but it assumes that the cognitive measures are normally distributed.
  • Confidence Interval Interpretation: The 95% confidence intervals are correctly applied, providing an estimate of the precision of the observed effects. However, the interpretation of these intervals should be cautious, particularly when making inferences about the efficacy of the interventions in the absence of p-values.
  • Subgroup Analysis Validity: The subgroup analyses via forest plots are a valuable addition, allowing for the exploration of heterogeneity of treatment effects. However, these analyses should be interpreted with caution due to the potential for increased Type I error when conducting multiple comparisons. The rationale for choosing these specific subgroups should be clearly justified.
  • Control for Confounders: It is not immediately clear from the figure whether the analyses have been adjusted for potential confounders. While the caption mentions 'baseline subgroups,' it is important that the analyses account for other factors that could influence cognitive change.
Communication
  • Clarity of Graphical Presentation: The graphical presentation is generally clear, with separate panels for each intervention and clear labeling of axes. However, the use of z-scores may be difficult for some readers to interpret, and the figure could benefit from a more explicit explanation of what a z-score represents in this context.
  • Forest Plot Readability: The forest plots are relatively standard in their presentation, but the sheer number of subgroups presented could be overwhelming. It may be beneficial to highlight the most relevant or statistically significant subgroups to aid in interpretation.
  • Caption Completeness: The caption provides a good overview of the figure's content but could be improved by briefly explaining the rationale for the subgroup analyses and the implications of the findings. Additionally, defining 'marginal differences' within the caption would enhance understanding.
  • Visual Clutter: There is a risk of visual clutter due to the amount of information presented. Simplifying the figure by focusing on fewer, key subgroups or using supplementary materials for less critical analyses could improve readability.
FIGURE 4 Three-year change in the episodic memory composite (A) and executive...
Full Caption

FIGURE 4 Three-year change in the episodic memory composite (A) and executive function composite (B) for the active and placebo multivitamin-mineral groups (plotted values: mean standardized (z) scores and 95% confidence intervals).

Figure/Table Image (Page 9)
FIGURE 4 Three-year change in the episodic memory composite (A) and executive function composite (B) for the active and placebo multivitamin-mineral groups (plotted values: mean standardized (z) scores and 95% confidence intervals).
First Reference in Text
MVM supplementation led to relative improvements both for memory (mean change z-score = 0.06, 95% CI: 0.002 to 0.13; P = .04) and for executive function (mean change z-score = 0.06, 95% CI: 0.01 to 0.11; P = .02) (Figure 4).
Description
Scientific Validity
Communication
TABLE 1 Distribution of baseline characteristics for COSMOS-Mind participants...
Full Caption

TABLE 1 Distribution of baseline characteristics for COSMOS-Mind participants by cocoa extract (CE) and multivitamin-mineral (MVM) treatment assignment

Figure/Table Image (Page 6)
TABLE 1 Distribution of baseline characteristics for COSMOS-Mind participants by cocoa extract (CE) and multivitamin-mineral (MVM) treatment assignment
First Reference in Text
Baseline characteristics and baseline cognitive test scores were well balanced across treatment groups (Table 1 and Table S1).
Description
  • Purpose of the Table: This table shows the characteristics of the participants at the start of the study, called 'baseline characteristics.' These characteristics include things like age, sex, race, education, health status, and lifestyle factors. The purpose of showing this information is to demonstrate that the different treatment groups were similar to each other at the beginning of the study. This is important because if the groups were very different at the start, it would be hard to tell if any changes seen later were due to the treatments or just because of these initial differences.
  • Treatment Groups: The participants were divided into four groups based on whether they received cocoa extract (CE), a multivitamin-mineral supplement (MVM), both, or neither. The table compares these groups to see if they were similar in terms of their baseline characteristics. For example, it checks if the average age or the proportion of men and women was roughly the same in each group. A 'placebo' is a dummy treatment that has no active effect, used here as a control to compare against the actual treatments. So, 'CE-Placebo, MVM-Placebo' means participants in this group got neither the cocoa extract nor the multivitamin.
  • Types of Characteristics Shown: The table includes a variety of characteristics, such as demographic information (age, sex, race, education), health measurements (body mass index, blood pressure), medical history (diabetes, cardiovascular disease), medication use, lifestyle factors (exercise, smoking, alcohol, chocolate intake), and cognitive function scores. These characteristics are important because they could potentially influence how a person responds to the treatments.
  • Statistical Measures: The table uses different statistical measures to describe the characteristics. For continuous variables like age or body mass index, it shows the 'mean (SD),' which stands for the average value and the standard deviation (a measure of how spread out the values are around the mean). For categorical variables like sex or smoking status, it shows the number and percentage of participants in each category (e.g., '914 (40.4)' for males, meaning 914 participants, or 40.4% of the total, were male). 'Median (IQR)' is used for variables that might not be normally distributed, where the median is the middle value, and the IQR (interquartile range) is the range between the 25th and 75th percentiles, giving an idea of the spread of the middle 50% of the data.
  • Importance of Balance: The table is meant to show that the treatment groups were 'well balanced,' meaning there were no major differences between the groups in any of the baseline characteristics. This balance is crucial for the validity of the study because it means that any differences in outcomes observed later are more likely to be due to the treatments themselves rather than pre-existing differences between the groups.
Scientific Validity
  • Comprehensive Baseline Assessment: The table provides a comprehensive assessment of a wide range of baseline characteristics, which is crucial for evaluating the comparability of the treatment groups. This thoroughness enhances the internal validity of the study by ensuring that potential confounders are identified and can be accounted for in the analysis.
  • Appropriate Statistical Reporting: The use of means, standard deviations, medians, and interquartile ranges, as well as counts and percentages, is appropriate for the types of data presented. However, the table lacks p-values, which would typically be used to formally assess the statistical significance of any differences between groups.
  • Relevance of Variables: The variables included in the table are relevant to the study's research questions and potential confounding factors. However, the rationale for including each variable could be more explicitly stated, particularly for those that may not be immediately obvious to the reader.
  • Potential for Residual Confounding: While the table demonstrates a good balance across many variables, there is always the potential for residual confounding due to unmeasured or unknown factors. The authors should acknowledge this limitation and discuss its potential impact on the study's findings.
Communication
  • Organization and Clarity: The table is well-organized, with clear headings and logical grouping of variables. The use of abbreviations is consistent, and the caption provides a good overview of the table's content. However, the table is quite dense, and the sheer amount of information presented could be overwhelming for some readers. It may be helpful to use bolding or shading to highlight key findings or to separate different categories of variables more distinctly.
  • Readability: The font size and formatting are generally readable, but the table spans multiple pages, which can make it difficult to follow. It might be beneficial to present the table as a supplementary file or to split it into smaller, more focused tables.
  • Accessibility for Non-Experts: The table is primarily aimed at a scientific audience and uses technical terms that may not be familiar to non-experts. While some terms are explained in the footnotes, others are not. Providing a brief glossary of terms or expanding the footnotes could improve accessibility.
  • Inclusion of Statistical Significance: The table would be strengthened by the inclusion of p-values to indicate the statistical significance of any differences between groups. This would provide a more objective assessment of the balance achieved by randomization and help readers interpret the importance of any observed differences.

4 DISCUSSION

Key Aspects

Strengths

Suggestions for Improvement

Non-Text Elements

FIGURE 3 Three-year change in global cognition composite for the active and...
Full Caption

FIGURE 3 Three-year change in global cognition composite for the active and placebo multivitamin-mineral groups by history of cardiovascular disease, which was based on self-report of transient ischemic attack, congestive heart failure, coronary artery bypass graft, angioplasty, or stent (plotted values: mean standardized (z) scores and 95% confidence intervals).

Figure/Table Image (Page 9)
FIGURE 3 Three-year change in global cognition composite for the active and placebo multivitamin-mineral groups by history of cardiovascular disease, which was based on self-report of transient ischemic attack, congestive heart failure, coronary artery bypass graft, angioplasty, or stent (plotted values: mean standardized (z) scores and 95% confidence intervals).
First Reference in Text
The MVM-treated CVD history subgroup had sustained increases in cognitive function after 2 years (Figure 3), while the placebo-treated CVD history subgroup showed cognitive decline after Year 1.
Description
  • Focus on Cognitive Change: This figure shows how a person's overall thinking and memory abilities, called 'global cognition,' changed over three years. It specifically looks at these changes in two groups of people: those who took a multivitamin-mineral supplement (active group) and those who took a pill with no active ingredients (placebo group).
  • Comparison Based on Cardiovascular Disease History: The figure divides each group (active and placebo) further into two subgroups based on whether they had a history of cardiovascular disease (CVD). Cardiovascular disease refers to conditions affecting the heart and blood vessels. The history is determined by participants' self-reporting of specific events like a transient ischemic attack (often called a 'mini-stroke'), congestive heart failure, or procedures like coronary artery bypass graft, angioplasty, or stent placement.
  • Standardized (z) Scores: The changes in cognition are shown using 'standardized (z) scores.' A z-score is a statistical measure that describes a value's relationship to the mean (average) of a group of values, measured in terms of standard deviations from the mean. If a z-score is zero, it indicates the data point's score is identical to the mean score. A positive z-score indicates the value is above the mean, while a negative z-score indicates it is below the mean. Here, it's used to show how much each individual's cognitive performance changed relative to their own starting point (baseline).
  • 95% Confidence Intervals: The '95% confidence intervals' indicate a range within which we are 95% certain that the true average change in cognition lies for each group. It's like saying, 'We're pretty sure the real average change is somewhere between these two values.' These are usually represented graphically as bars or lines extending from a plotted point, giving a visual sense of the precision of the estimate.
  • Plotted Values: The figure plots the average (mean) change in cognitive scores for each group over the three years. This allows for a visual comparison of how cognition changed in the active group versus the placebo group, and how this differed between those with and without a history of cardiovascular disease.
Scientific Validity
  • Self-Reported CVD History: The reliance on self-reported cardiovascular disease history is a potential limitation. Self-reports can be subject to recall bias or misinterpretation, which could affect the accuracy of subgroup categorization. It would be preferable to have this confirmed by medical records.
  • Appropriateness of Outcome Measure: Using a global cognition composite is suitable for a broad assessment of cognitive function. However, the specific domains of cognition most affected by MVM supplementation in those with CVD history are not discernible from this figure alone. Further breakdown into domain-specific scores could provide more nuanced insights.
  • Statistical Analysis: The use of z-scores and 95% confidence intervals is appropriate for presenting the data. However, without p-values or tests for statistical significance, it is difficult to determine the certainty of the observed differences between groups.
  • Potential Confounding Factors: While the study attempts to isolate the effect of CVD history, other factors could confound the relationship between MVM supplementation, CVD history, and cognitive change. These potential confounders should be addressed in the statistical analysis.
Communication
  • Clarity of Grouping: The figure clearly distinguishes between active and placebo groups and further stratifies by CVD history, which is crucial for understanding the study's findings. However, the specific criteria for 'CVD history' could be more prominently displayed on the figure itself for better clarity.
  • Visual Representation: The use of a line graph to depict changes over time is effective for showing trends. However, the figure could be enhanced by including markers for each year to clearly indicate the time points at which measurements were taken.
  • Caption Detail: The caption adequately describes the figure but could benefit from a brief explanation of why CVD history was chosen as a stratifying variable. Additionally, including a statement about the statistical significance of the observed differences would strengthen the interpretation.
  • Accessibility for Non-Experts: While the figure is generally understandable, the concept of z-scores might be challenging for some readers. Providing a lay-friendly explanation of z-scores in the caption or a supplementary note could improve accessibility.

RESEARCH IN CONTEXT

Key Aspects

Strengths

Suggestions for Improvement

↑ Back to Top