Chocolate Consumption on Cardiovascular Disease

Written by Zachary Wenger

July 27, 2021

Disclaimer: I’m currently a college student and dietetic major, not a registered dietitian. None of this should be taken as medical advice.

Within the field of nutrition science, chocolate receives a substantial amount of popularity for its cardioprotective properties (1). Heck, in addition, the more chocolate that I eat, I should expect a rise in cognition such that I should be winning a Nobel prize in the near future (2). What’s not to love about chocolate?

If you were unable to tell, I was being facetious to the Nobel prize finding. Researchers found a statistically significant, linear association (r=0.791, P<0.0001), between chocolate consumption and the number of Nobel prizes in 23 countries. It is crucial to emphasize that this is an ecological study that doesn’t adjust for any confounders. A few researchers took issue with the over-extrapolation of the finding and wanted to demonstrate a comical, yet critical point. These researchers conducted their own ecological study and found an even stronger correlation between the number of IKEA stores and Nobel prizes (r= 0.82; P< 0.0001) (3). In regards to methodological rigor, and at any attempt to produce a causal inference, this finding would be at the bottom of the barrel.

All jokes aside, there actually is quite a bit of research investigating chocolate consumption on cardiovascular disease (CVD). Is the buzz around chocolate consumption and its impact on CVD warranted or considerably overstated? We’ll examine the prospective cohort data extensively, as well as investigate biological plausibility, if the effect lives up to the hype. Sit back, grab yourself some dark chocolate, and enjoy!

Chocolate and Cardiovascular Disease:

Firstly, let’s begin with the relationship between chocolate consumption and CVD. The most recent meta-analysis and dose-response analysis of prospective cohort studies investigating this topic revealed that chocolate consumption statistically significantly reduced risk for CVD by 11% (RR, 0.89; 95% CI: 0.849–0.932). Maximal reductions in risk occurred at 45 grams/week, which is equivalent to a single chocolate square per day or a single chocolate snack bar per week (4). Furthermore, the reduction in risk transitions to null, once chocolate intake reaches 100 grams/week.

It’s important to always keep in mind that the strength of any given meta-analysis is only as strong as the studies included. In other words, looking through the individual studies included in a given meta-analysis is a salient aspect of an effective critical appraisal. Upon doing so, I noticed that two of the studies included in the meta-analysis reported chocolate consumption as a dichotomous variable. For those unfamiliar, a dichotomous variable is a type of variable that only takes on two possible values. Mink, 2007 was one of the papers that did this (5). Mink and colleagues were investigating the relationship between flavonoid intake and a collection of cardiovascular outcomes. These researchers further broke it down by investigating individual flavonoid rich foods, which included wine, apples, pears, orange juice, bran, and chocolate, on CVD, coronary heart disease (CHD), and stroke mortality. When investigating low to high servings/week of chocolate, researchers observed a statistically significant 22% reduction in stroke mortality (RR, 0.78; 95% CI: 0.65–0.94), while the age and energy adjustment model were applied. However, the risk lost statistical significance after the implementation of their multivariate adjustment model (RR, 0.85; 95% CI: 0.70–1.03).

The predominant methodological issue of this paper arises when you take into consideration the lack of contrast within the intake ranges. If you take a glance at table 7 (seen above), the intake range in question was 0 servings of chocolate/week, relative to >0 servings of chocolate/week. For clarity, I am not sure why, or how, this cohort was included in the dose-response curve. Specifically, the individual cohort reported on servings/week, while the dose-response curve required grams/week. However, one could make the argument that the researchers could simply convert a standardized serving of chocolate into grams. While that is technically true, recall that the value in question is “>0 servings/week.” >0 servings/week has no ceiling and could hypothetically fit anywhere over 0 in the dose-response curve. In other words, there wasn’t a specific dose that was isolated and investigated, to plot on the dose-response curve.

This occurs again with Lewis, 2011, which is inserted above, which also reported chocolate consumption as a dichotomous variable. This time around, the intake range in question was <1 serving of chocolate/week, relative to ≥1 serving of chocolate/week (6). Once again, this introduces interpretive challenges. We could convert servings/week into grams/week, however, within the dose-response curve, would this hazard ratio fall under 1 serving of chocolate at 30 grams, 2 servings at 60 grams, 3 servings at 90 grams, etc? Again, ≥1 serving of chocolate/week lacks a ceiling and reporting chocolate consumption as a dichotomous variable, can only truly be interpreted as a “eat/don’t eat” type of scenario. Nonetheless, if that was the academic question that researchers had, interpreting this as just an individual study could be justified. However, attempting to plot this on a dose-response curve, without any mention of a median range, is not sound methodology.

In addition, another unfortunate methodological limitation within both of these studies is the lack of adjustment for any dietary covariate, besides total energy intake. Including adjustments for dietary covariates can assist with teasing out the independent effects of the exposure, which in this instance is chocolate. While both of these cohorts presented reductions in CVD risk, I would be hesitant to suggest that these are anything to hang your hat on. Fortunately, these two studies aren’t representative of the literature on chocolate consumption and CVD. 

The ideal way to approach the topic would be to investigate the well designed cohorts, individually. For example, investigating cohorts that present a wide contract in chocolate intake, large sample sizes, reasonable event rates, along with a sound adjustment model that includes critical covariates that may impact disease risk. Could it be that once limited to just those studies, the beneficial effect of chocolate on CVD risk dissipates? Alternatively, could it be that the risk reduction sustains, even while limited to well designed studies? Let’s find out.

Firstly, let’s examine Larsson, 2012, which investigated chocolate consumption on stroke incidence in the Cohort of Swedish Men (7). By reference to table 2, which is included above, notice the wider contrast in chocolate intake. We’re able to observe differences in chocolate consumption of up to 62.9 grams/week, relative to 0 grams/week. In addition, notice how the researchers wisely controlled for critically important dietary covariates, other than just total energy intake, such as alcohol, coffee, tea, red meat, processed meat, fish, fruits, and vegetables. These characteristics can assist in teasing out the independent effects of chocolate. Furthermore, when investigating the risk of total stroke between 0 grams/week, relative to those that consumed 12.5 grams/week, as well as 38.4 grams/week, there was no statistically significant difference in risk (RR, 0.94; 95% CI: 0.84–1.05 and RR, 0.95; 95% CI: 0.83–1.08, respectively). However, when the intake subgroup was broadened to 62.9 grams/week, there was a statistically significant 17% reduction in stroke risk, relative to non-consumers of chocolate (RR, 0.83; 95% CI: 0.70–0.99).

Next, let’s investigate Buijsse, 2010 (8). Inserted above is table 3, which provides the relative risk of chocolate consumption on the risk of myocardial infarction (MI), stroke, and MI + stroke, from the Potsdam arm of the EPIC cohort. The grams of chocolate consumption for Q1 is 11.9 grams/week (reference), 13.3 grams/week for Q2, 23.1 grams/week for Q3, and 52.5 grams/week for Q4. At the intake range of 52.5 grams/week, we observe a massive 39% reduction in MI + stroke risk, relative to 11.9 grams/week (RR, 0.61; 95% CI: 0.44–0.87). This reduction in risk was in the presence of an adjustment model including dietary covariates such as fruit, vegetables, red meat, processed meat, dairy, coffee, tea, and cereal fiber. However, a noteworthy shortcoming of this study was that the event rate was magnitudes lower than Larsson, 2012, which was the previous study that we discussed, along with the upcoming research that I will present later in the article. 

Dong, 2017 from the Japan Public Health Center-based Prospective Study was another well powered cohort (9). When comparing 37.5 grams/week to 0 grams/week, the association on total stroke risk was null for men (HR, 0.94; 95% CI: 0.80–1.10), though a statistically significant 16% reduction in women (HR, 0.84; 95% CI: 0.71–0.99). Dietary covariates that were adjusted for include: green tea, coffee, fish/seafood, meat, fruits, soy foods, and vegetables. If there are a lot of numbers and information that you are attempting to retain, don’t worry, we will piece this all together at the end. 

Larsson, 2011 is another excellent prospective cohort study from the Swedish Mammography Cohort (10). The multivariable adjustment model for dietary components include: total calories, alcohol, coffee, tea, fresh red meat, processed red meat, fish, fruits, and vegetables. Chocolate consumption is broken up into grams/week, in which the median values are 8.8 grams/week for Q1 (reference), 14 grams/week for Q2, 28.5 grams/week for Q3, and 66.5 for Q4. When investigating Q4, in which the median value was 66.5 grams/week, we observe a statistically significant 20% reduction in total stroke risk, relative to Q1 of 8.8 grams/week (RR, 0.80; 95% CI: 0.66–0.99). Chocolate, my friend, you’re glorious.

Let’s now take a look at Larsson, 2016, which includes the Cohort of Swedish Men and the Swedish Mammography Cohort (11). Susanna Larsson is a clever researcher and has published quality prospective cohort studies on chocolate consumption on CVD. Hence, why this is now her third feature in this article. This study is unique amongst her other publications on chocolate and CVD, as this is broken up into serving quartiles, rather than grams quartiles. Nevertheless, we still observe a wide contrast in chocolate consumption, and while investigating risk in those that consume ≥3-4 servings/week, we observe a 13% statistically significant reduction in risk of MI, relative to 0 servings/month (RR, 0.87; 95% CI: 0.77–0.98). Dietary covariates that were utilized in model 2 of the multivariable adjustment model include total energy, alcohol, processed meat, as well as fruits and vegetables. 

Staying on the topic of investigating servings/frequency, we can also look through Greenberg, 2018, from the Women’s Health Initiative cohort, with the outcome including the risk of CHD, stroke, or both (40). The dietary covariates included in the adjustment model include the Alternative Healthy Eating Index, non chocolate daily energy intake, and alcohol consumption. This cohort is unique amongst the rest of the studies that we have gone through, as it found non-significant results for all of the subgroups investigated (utilizing the full model). A couple of the confidence intervals nearly overlapped 1. Nonetheless, while this study appears to be an outlier, relative to other prospective cohorts, this will be included in our final analysis of the research. 

With the most recent publication within our analysis, let’s investigate Ho, 2021, from the MVP cohort, with the outcome including fatal and nonfatal coronary artery disease (CAD) (41). Despite there being an incredibly short follow-up time of 3.2 years, researchers still found statistically significant reductions in risk. While utilizing model 4 of the adjustment model, which includes an adjustment for the DASH score and alcohol consumption, while exploring high vs low intakes of chocolate, there was a statistically significant 11% reduction in fatal and nonfatal CAD at ≥5 servings/week, relative to those consuming <1 serving/month (HR, 0.89; 95% CI: 0.83–0.95).

Lastly, we can investigate the Cohort of Swedish Men once more, with Steinhaus, 2017, as we examine the relationship between chocolate consumption and heart failure (HF) hospitalization or mortality (12). In addition, this would appear to be the largest intake spread that we have observed yet. While utilizing model 2 of the adjustment model, which includes an adjustment for the DASH diet component score, there was a statistically significant 18% reduction in HF from 3-6 servings/week, relative to non-consumers (HR, 0.82; 95% CI: 0.68–0.99). However, when examining ≥1 servings/day, relative to non-consumers, there were no statistically significant differences in risk (HR, 1.10; 95% CI: 0.84–1.45). In other words, the protective effect that chocolate had on HF hospitalization or mortality, of up to 3-6 servings/week, was not detected at ≥1 servings/day, which appears to be a J-shaped curve.

Healthy User Bias:

Now, as a critical aside, if you’re familiar with attempts to discredit observational data, you are in all likelihood aware of the healthy user bias. For those unfamiliar, the healthy user bias is characterized as healthy habits performed by individuals that distort the outcome to an extent that we are uncertain if the exposure in question is creating the effect or the healthy habits. For example, let’s attempt to invoke the healthy user bias in the case of fruit consumption on CVD risk. Let’s imagine in our hypothetical prospective cohort, that the results find a 30% reduction in CVD risk, while comparing the highest quartile of fruit consumers, to the lowest. Now, we could attempt to explain the findings by pointing to the healthy habits of those that consume large quantities of fruit, such as frequent exercise, less smoking, less alcohol consumption, etc, and conclude that those factors are what caused the reduction risk for disease and not the fruit.

Firstly, researchers are well aware of the healthy user bias, and statistically adjust for these critical covariates. In other words, the confounders that we previously mentioned, including exercise, smoking, and alcohol consumption, are often statistically adjusted for, in order for a potential imbalance in risk mediators, to be accounted for and minimized. Nonetheless, even if these adjustments weren’t an option, can we even appeal to the healthy user bias in the case of chocolate? By examining table 1 of Larsson 2012 (7), which presents the baseline characteristics of the cohort, we notice that among those that consume more chocolate, consume more fruits, vegetables, tea, and have lower rates of hypertension. However, those in the highest quartile of chocolate consumption also consume more alcohol, red meat, processed meat, total calories, less fish, and also participate in less physical activity, relative to the lowest quartile of chocolate consumption. In other words, healthy habits aren’t overwhelmingly in favor of high chocolate consumers and you can make perfectly reasonable arguments in the opposite direction.

To reiterate, in well designed prospective cohort studies, established covariates that impact disease risk are generally statistically adjusted for. In addition, in the case with chocolate, even if we were to ignore the adjustments or provide an appeal to distrust them, high chocolate consumption isn’t staggeringly associated with healthier behaviors. In fact, chocolate consumption seems to be a correlate for many unhealthy behaviors.

Interpretations of the Research:

Recall that before examining the individual cohorts, the question on the table was that if limited to well designed prospective cohorts, would the risk reduction of chocolate on CVD remain consistent, relative to the effects that we observed within the dose-response curve? The answer to this question would overwhelmingly be yes. In an attempt to best interpret the data, I created a forest plot to present all of the data that we examined, on chocolate consumption and CVD, in primary prevention. In addition, I created different subgroups to present the data based on if the reporting was in grams or servings. For clarity, the dose-response curve that we inspected earlier, included 14 publications. I have only included 6 of those publications. My exclusion criteria consisted of studies that didn’t adjust for any dietary covariates other than total energy and/or alcohol (14,15,17,18) presented data in secondary prevention (13), presented data on cocoa intake/not chocolate (16), or presented data as a dichotomous variable that wouldn’t be pertinent for investigating dose (5,6). 

A common misconception with interpreting meta-analytic summations is that more always means better. This would only apply if all else was held equal, though that’s just unrealistic. There are a handful of explanations for this, which could include papers that lack rigorous statistical adjustments and power, over-adjusting by controlling for mediators, lack of an acceptable contrast in intake, or quite simply, that the study is simply not appropriate for the question that you are attempting to answer. The point being, while interpreting meta-analyses, it is entirely possible to have a greater degree of confidence within a relationship, with fewer studies included. 

By examining the forest plot, all but two subgroups found statistically significant reductions in CVD risk, with the intake ranges of chocolate consumption provided. I included subgroups for both servings and grams, as different studies reported on different measurements. My forest plot largely echoed the dose-response curve that we examined originally. The only subgroups that generated non-significant findings were <1 serving/month to 1-3 servings/month (RR, 0.95; 95% CI: 0.87–1.03), as well as <1 serving/month to ≥1 serving/day (RR, 1.10; 95% CI: 0.84–1.45). This is vastly similar to the J-shaped curve that we observed in the dose-response curve. With the two subgroups that provided us with non-significant findings, it could be that further research could strengthen the association, or that further research could weaken the association. Nonetheless, in the aggregate, the research is largely in favor of chocolate consumption on CVD related outcomes.

An important question that you may have is about the type of chocolate that was consumed in these cohorts. Four out of the eight cohorts were carried out in Sweden, and the researchers report that approximately 90% of chocolate consumption was milk chocolate in the 1990s (7,10,11,12). This reporting was derived from Kraft Foods Sverige, a large snack company in Sweden (19). Unfortunately, other than the country wide reporting, the type of chocolate in these Swedish cohorts weren’t directly reported on. However, with Larsson, 2016 (one of the Swedish cohorts), the researchers reported: “Although we had no information on the proportion of dark versus milk chocolate intake in the whole study population, in a subsample of participants who had completed an extensive FFQ in 2010, 54% of participants reported that they consumed more dark chocolate than milk chocolate (unpublished data)” (11). This would appear odd, as the original quote from Larsson and Kraft Foods, would present us with an exceedingly different figure. We would expect that milk chocolate be the primary source of chocolate, yet the opposite was true. Even if we were to grant that there could have been a population-wide shift with the types of chocolate consumed in Sweden, which isn’t reported anywhere from 1990-2010, it would be incredibly unlikely that shift would be that drastic. There could be a few explanations here. It could be that the number of study participants were too low in the unpublished reporting and we are presented with outliers consuming primarily dark chocolate. It could be that the reporting wasn’t from Sweden, though that would be odd as Susanna Larsson is a Swedish researcher and the quote implies that the subsample was derived from the same cohorts (Cohort of Swedish Men and/or the Swedish Mammography Cohort). My overwhelming suspicion is that the reporting from Kraft Foods Sverige, quite simply didn’t represent the majority of chocolate consumed in Sweden at the time.

Moving on, the study participants in the Potsdam arm of the EPIC study reported that 57% consumed milk chocolate, 24% consumed dark chocolate, only 2% consumed white chocolate, and 17% the type of chocolate was unspecified (8). This was the only study that reported on the breakdown of chocolate consumption. The study participants from the Japan Public Health Center-based Prospective Study didn’t report on the type of chocolate consumed (9). The Women’s Health Initiative cohort reported solely on chocolate candy or candy bars, so the type of chocolate was unspecified (40). Lastly and similarly, the MVP cohort did not differentiate between white, milk, or dark chocolate consumption (41).

Evidently, the type of chocolate consumed in the cohorts is quite the mystery. My overarching position on this is that we don’t know which type of chocolate provided the benefits in the prospective cohort data, other than the EPIC cohort, which provided us with the chocolate consumption breakdown. Future research is desperately needed to determine if the type of chocolate has differential effects on CVD risk (e.g., dark vs milk vs white chocolate on CVD risk).

Biological Plausibility:

Now that we are aware that the finding is consistent and robust, we can now go over explanations as to why this may be occurring. To start, it is important to touch on the fatty acid composition of chocolate, as this often creates a bit of confusion. Researchers in Turkey compared 20 different trademarks of chocolate from local markets, with the objective to determine the fatty acid composition and cholesterol content of different chocolates (20). The researchers reported that 69.95% of total fatty acids were composed of saturated fatty acids, with the predominant form of saturated fat being stearic acid (39.33 + 5.25%). How could a food that is this rich in saturated fat be cardioprotective? 

An explanation to this is that stearic acid appears to stand out, as relative to other saturated fatty acids, it presents neutral effects on blood lipids (21,22). Now, I don’t hold the position that this is an explanation to why chocolate appears incredible for reducing CVD risk. However, it is important to emphasize that chocolate is an outlier, relative to other foods rich in saturated fat, as a direct result of the primary saturated fatty acid, stearic acid.

A mechanism that seems more plausible is the polyphenolic content of chocolate. Polyphenols are naturally occurring compounds that are found in plants that are thought to have beneficial effects on human health. Cocoa beans specifically, are an incredibly rich source of polyphenols, though the polyphenolic profile in cocoa may vary, depending on several factors, such as the plant variety, geographical location, and ripeness of the bean (23). 

While a wide array of polyphenolic compounds are found in cocoa beans, cocoa powder is incredibly rich in a subcategory of polyphenols known as flavonoids. Primarily, the flavonoids, catechin, epicatechin, and proanthocyanidins (24). The first table inserted above, illustrates that not all forms of chocolate are created equal, regarding their total polyphenol content or flavanol content. Dark chocolate is formulated with a higher proportion of cocoa, relative to milk chocolate, which explains why dark chocolate contains greater amounts of flavonoids and total polyphenols. While milk chocolate does contain less flavonoids, relative to dark chocolate, it still contains a considerable amount, relative to other foods, which is listed in the second table above (24). 

Now, there are a collection of proposed mechanisms to why these polyphenols and flavonoids appear to reduce risk for CVD. Firstly, let’s discuss cocoa flavonoids on blood pressure. High blood pressure is a leading risk factor for CVD, which is the leading cause of mortality worldwide (25). A collection of cells that play a large role in blood pressure homeostasis is the endothelium, a single layer of endothelial cells. For those unfamiliar, the endothelium is a thin membrane that lines the inside of the heart and blood vessels. Furthermore, the endothelium releases vasodilating compounds, which includes a molecule known as nitric oxide (NO) (26). Genetically enhanced NO signaling was associated with reduced risks of coronary heart disease (OR, 0.37; 95% CI: 0.31–0.45), peripheral arterial disease (OR, 0.42; 95% CI: 0.26–0.68), and stroke (OR, 0.53; 95% CI: 0.37–0.76) (27). Inversely, genetically restricted NO signaling was associated with a 22.8 mm Hg (95% CI: 11.7–33.9) higher systolic blood pressure, 9.71 mm Hg (95% CI: 3.52, 15.90) higher diastolic blood pressure, and a higher risk of coronary heart disease (OR, 3.03; 95% CI: 1.29–7.12) (27). Needless to say, the endothelium’s role in releasing vasodilating compounds, such as NO, and the role that it plays in the heart and blood vessels, is incredibly important.

NO is synthesized from L-arginine, an amino acid, by the enzyme NO synthase (NOS) (28). In vitro, flavonoids are able to activate NOS (29). Cocoa flavonoids are hypothesized to lower blood pressure by the formation of NO by the endothelium, which promotes vasodilation and consequently lowers blood pressure. Fortunately, we are able to examine well designed randomized controlled trials to investigate the question: does cocoa polyphenols reduce blood pressure?

As a secondary outcome, researchers out of Italy aimed to investigate changes in blood pressure as a function of cocoa polyphenols (30). Study participants were randomized to consume a dairy-based cocoa drink containing cocoa flavanols at high (HF: ~990 mg of flavanols per drink), intermediate (IF: ~520 mg of flavanols per drink), or low levels (LF: ~45 mg of flavanols per drink), once a day for 8 weeks. Relative to baseline, there was a statistically significant reduction in systolic and diastolic blood pressures of the HF (systolic: −10.0±3.1 mm Hg, P<0.0001; diastolic: −4.8±1.8 mm Hg, P<0.0001) and IF (systolic: −8.2±3.5 mm Hg, P<0.0001; diastolic: −3.4±2.0 mm Hg, P<0.0001) subjects. However, not in the study participants in the LF group (systolic: −1.4±5.4 mm Hg, P=0.16; diastolic: −0.9±3.4 mm Hg, P=0.14).

In the latest Cochrane analysis of randomized controlled trials (RCTs) on the effects of cocoa consumption on blood pressure, detected modest, though statistically significant, reductions in both systolic (-1.76, 95% CI: -3.09 mmHg to -0.43 mmHg) and diastolic blood pressure (-1.76, 95% CI: -2.57 to -0.94) (31). The heterogeneity was exceptionally high in their findings and the researchers attempted to seek out answers with sensitivity analyses. However, the sensitivity analyses didn’t adequately explain it, as they conclude: “While we investigated heterogeneity in several subgroup analyses, we could not fully explain the variations in effect of cocoa on blood pressure. Continuing high levels of heterogeneity within subgroup analyses suggest that there may be a combination of factors, or additional ones beyond those we considered.”

Nonetheless, the results suggest that chocolate may reduce risk for cardiovascular disease as a function of modest, yet statistically significant, reductions in blood pressure. In addition, there is also the effect of cocoa polyphenols on LDL oxidation. In a randomized crossover design, subjects were randomized to nearly identical diets, one being a standard american diet, which was the control, while the other was a standard american diet + 22g of cocoa powder and 16g of dark chocolate, which was the intervention (32). In other words, these two diets were incredibly similar, other than differences in the polyphenol content, which was overwhelmingly in favor of the cocoa powder + dark chocolate diet. The primary finding was differential effects on lag time to LDL oxidation. For those unfamiliar, lag time denotes the amount of time it takes for LDL particles to undergo oxidation. The lag time to LDL oxidation may be used to represent the serum total antioxidant capacity (33). In the current study, lag time was statistically significantly ~8% greater with the cocoa powder + dark chocolate diet, relative to the control diet. As follows, the serum total antioxidant capacity was ~4.2% higher when subjects consumed the cocoa powder + dark chocolate diet, relative to when they consumed the control diet. The mechanism for this is that flavonoids are absorbed and can then bind to LDL particles, which can inhibit LDL oxidation. This mechanism has been specifically demonstrated with red wine, which is also rich in polyphenols (34). 

The finding was replicated with another study design. Study participants were randomized to either 12 grams of sugar/day (control group) or 26 grams of cocoa powder/day + 12 g sugar/day (intervention group) for 12 weeks. The study participants were either normocholesterolemic or mildly hypercholesterolemic. Once again, in the intervention group, which included the cocoa powder, there was a 9.4% prolongation in lag time to LDL oxidation (35). 

While we are aware that cocoa flavonoids increase the lag time to LDL oxidation and modestly reduce blood pressure, which could explain the reductions that we observe in cardiovascular disease, could we directly investigate flavonoid intake on the outcome itself? Maybe these mechanisms work in concert with one another, to assist in reducing risk for cardiovascular disease. Fortunately, we most certainly can. In the latest dose-response meta-analysis of prospective cohort studies, we are able to examine the outcomes of all-cause mortality (ACM) (left) and CVD mortality (right), as a function of the flavonoid intake (36). For ACM, we observe a maximal risk reduction of 23% at 400 mg/day of flavonoids (RR, 0.77; 95% CI: 0.61–0.97). For cardiovascular disease mortality, we observe a maximal risk reduction of 26% at 700 mg/day of flavonoids (RR, 0.74; 95% CI: 0.56–0.97). These findings suggest that the reduction in cardiovascular disease that we extensively examined with chocolate consumption, could be explained by the flavonoid content of cocoa. 

However, I did notice that the inclusion criteria allowed for studies with no dietary adjustments, which is problematic for attempting to tease out the independent effects of flavonoids. By examining one that does contain adjustments for dietary covariates (low fruit and vegetable intake), Ivey, 2015, which investigated flavonoid intake in women aged >75 years old, on ACM, CVD, and cancer mortality (37), we observe a massive 68% reduction in cardiovascular disease mortality, which was indeed statistically significant, when comparing ≥788 mg/day to <525 mg/day (HR: 0.32; 95% CI: 0.16–0.61).

In addition, Ponzo, 2015 is another paper to touch on, as it contains wide variations in flavonoid intakes, as well as adjusting for crucial dietary covariates, such as fiber, saturated fat, and alcohol intake (38). The first tertile contained a median intake of 89 mg/day (reference), the second tertile at 251.4 mg/day, and the third tertile at 532.3 mg/day. While the effects were null for cardiovascular disease mortality (HR, 0.83; (95% CI: 0.46–1.51) and ACM (HR, 0.78; 95% CI: 0.55–1.13), when comparing 532.3 mg/day of flavonoids to 89 mg/day, there was a statistically significant 54% reduction in non-fatal cardiovascular events (HR, 0.46; 95% CI: 0.28–0.75). 

Lastly, let’s examine Zamora-Ros, 2013 (39). By investigating 241.2-329.8 mg/day, relative to <136.2 mg/day, we observe a statistically significant 29% reduction in ACM (HR, 0.71; 95% CI: 0.50–0.99). Risk beyond that intake range was null in this cohort. The adjustment model for dietary covariates included alcohol, total energy, vitamin C, and fiber intake.

Conclusions:

In conclusion, I’m willing to transition to the new phrase of “a chocolate square a day keeps the doctor away.” In all seriousness, the inverse relationship between chocolate consumption and cardiovascular disease risk is profoundly consistent, even while limited to well designed cohorts. There were a handful of cohort studies that were likely best excluded from the original dose-response curve investigating the relationship between chocolate consumption and cardiovascular disease. However, the direction of effect remained consistent in the updated forest plot that I created. In other words, while there may be some discrepancy with the inclusion criteria, the take home message generally remains the same.

The dose of chocolate is also incredibly important, as the relationship is not linear. The finding shouldn’t be interpreted as the more chocolate consumed, the lower the risk of cardiovascular disease. I’ll just quote the authors from the original dose-response curve, as they sum it up best: “Meanwhile, we predict that this non-linear association also exists for other types of CVD if there are enough data available, because higher levels of chocolate consumption may negate the health benefits and induce adverse effects along with high-level sugar consumption (4).” Maximal risk reductions in their meta-analysis were observed with 45g/week, which isn’t a lot of chocolate. For context, a full sized Hershey’s chocolate bar is 43 grams.

Nonetheless, my goal with this article was to present where the data currently stands on chocolate consumption in relation to cardiovascular disease. In addition, provide details on the individual cohorts that presented us with the findings. Future research desperately needs to differentiate between the different types of chocolate on CVD risk. We can hypothesize that it is likely a mixture of both, though future research can provide us with a greater deal of confidence. Evidently, this article was longer than my others, and I hope that you all found it interesting. I’m not sure about you all, though I could really go for some chocolate right now. Take care! 

References: 

  1. Fernández-Murga, L et al. “The impact of chocolate on cardiovascular health.” Maturitas vol. 69,4 (2011): 312-21. doi:10.1016/j.maturitas.2011.05.011
  2. Messerli, Franz H. “Chocolate consumption, cognitive function, and Nobel laureates.” The New England journal of medicine vol. 367,16 (2012): 1562-4. doi:10.1056/NEJMon1211064
  3. Maurage, Pierre et al. “Does chocolate consumption really boost Nobel Award chances? The peril of over-interpreting correlations in health studies.” The Journal of nutrition vol. 143,6 (2013): 931-3. doi:10.3945/jn.113.174813
  4. Ren, Yongcheng et al. “Chocolate consumption and risk of cardiovascular diseases: a meta-analysis of prospective studies.” Heart (British Cardiac Society) vol. 105,1 (2019): 49-55. doi:10.1136/heartjnl-2018-313131
  5. Mink, Pamela J et al. “Flavonoid intake and cardiovascular disease mortality: a prospective study in postmenopausal women.” The American journal of clinical nutrition vol. 85,3 (2007): 895-909. doi:10.1093/ajcn/85.3.895
  6. Lewis, Joshua R et al. “Habitual chocolate intake and vascular disease: a prospective study of clinical outcomes in older women.” Archives of internal medicine vol. 170,20 (2010): 1857-8. doi:10.1001/archinternmed.2010.396
  7. Larsson, Susanna C et al. “Chocolate consumption and risk of stroke: a prospective cohort of men and meta-analysis.” Neurology vol. 79,12 (2012): 1223-9. doi:10.1212/WNL.0b013e31826aacfa
  8. Buijsse, Brian et al. “Chocolate consumption in relation to blood pressure and risk of cardiovascular disease in German adults.” European heart journal vol. 31,13 (2010): 1616-23. doi:10.1093/eurheartj/ehq068
  9. Dong, Jia-Yi et al. “Chocolate consumption and risk of stroke among men and women: A large population-based, prospective cohort study.” Atherosclerosis vol. 260 (2017): 8-12. doi:10.1016/j.atherosclerosis.2017.03.004
  10. Larsson, Susanna C et al. “Chocolate consumption and risk of stroke in women.” Journal of the American College of Cardiology vol. 58,17 (2011): 1828-9. doi:10.1016/j.jacc.2011.07.023
  11. Larsson, Susanna C et al. “Chocolate consumption and risk of myocardial infarction: a prospective study and meta-analysis.” Heart (British Cardiac Society) vol. 102,13 (2016): 1017-22. doi:10.1136/heartjnl-2015-309203
  12. Steinhaus, Daniel A et al. “Chocolate intake and incidence of heart failure: Findings from the Cohort of Swedish Men.” American heart journal vol. 183 (2017): 18-23. doi:10.1016/j.ahj.2016.10.002
  13. Janszky, I et al. “Chocolate consumption and mortality following a first acute myocardial infarction: the Stockholm Heart Epidemiology Program.” Journal of internal medicine vol. 266,3 (2009): 248-57. doi:10.1111/j.1365-2796.2009.02088.x
  14. Kwok, Chun Shing et al. “Habitual chocolate consumption and risk of cardiovascular disease among healthy men and women.” Heart (British Cardiac Society) vol. 101,16 (2015): 1279-87. doi:10.1136/heartjnl-2014-307050
  15. Mostofsky, Elizabeth et al. “Chocolate intake and incidence of heart failure: a population-based prospective study of middle-aged and elderly women.” Circulation. Heart failure vol. 3,5 (2010): 612-6. doi:10.1161/CIRCHEARTFAILURE.110.944025
  16. Buijsse, Brian et al. “Cocoa intake, blood pressure, and cardiovascular mortality: the Zutphen Elderly Study.” Archives of internal medicine vol. 166,4 (2006): 411-7. doi:10.1001/archinte.166.4.411
  17. Kwok, C S et al. “Habitual chocolate consumption and the risk of incident heart failure among healthy men and women.” Nutrition, metabolism, and cardiovascular diseases : NMCD vol. 26,8 (2016): 722-34. doi:10.1016/j.numecd.2016.01.003
  18. Petrone, Andrew B et al. “Chocolate consumption and risk of heart failure in the Physicians’ Health Study.” European journal of heart failure vol. 16,12 (2014): 1372-6. doi:10.1002/ejhf.180
  19. Janszky, I et al. “Chocolate consumption and mortality following a first acute myocardial infarction: the Stockholm Heart Epidemiology Program.” Journal of internal medicine vol. 266,3 (2009): 248-57. doi:10.1111/j.1365-2796.2009.02088.x
  20. P. G. Ergönül , B. Ergönül & A. K. Seçkin (2010) Cholesterol content and fatty acid profile of chocolates mostly consumed in Turkey Contenido de colesterol y perfil de ácidos grasos de los chocolates más consumidos en Turquía, CyTA – Journal of Food, 8:1, 73-78, DOI: 10.1080/19476330903205033
  21. Clarke, R et al. “Dietary lipids and blood cholesterol: quantitative meta-analysis of metabolic ward studies.” BMJ (Clinical research ed.) vol. 314,7074 (1997): 112-7. doi:10.1136/bmj.314.7074.112
  22. Hunter, J Edward et al. “Cardiovascular disease risk of dietary stearic acid compared with trans, other saturated, and unsaturated fatty acids: a systematic review.” The American journal of clinical nutrition vol. 91,1 (2010): 46-63. doi:10.3945/ajcn.2009.27661
  23. Urbańska, Bogumiła, and Jolanta Kowalska. “Comparison of the Total Polyphenol Content and Antioxidant Activity of Chocolate Obtained from Roasted and Unroasted Cocoa Beans from Different Regions of the World.” Antioxidants (Basel, Switzerland) vol. 8,8 283. 6 Aug. 2019, doi:10.3390/antiox8080283
  24. Katz, David L et al. “Cocoa and chocolate in human health and disease.” Antioxidants & redox signaling vol. 15,10 (2011): 2779-811. doi:10.1089/ars.2010.3697
  25. “Cardiovascular Diseases.” World Health Organization, World Health Organization.
  26. Sandoo, Aamer et al. “The endothelium and its role in regulating vascular tone.” The open cardiovascular medicine journal vol. 4 302-12. 23 Dec. 2010, doi:10.2174/1874192401004010302
  27. Emdin, Connor A et al. “Phenotypic Consequences of a Genetic Predisposition to Enhanced Nitric Oxide Signaling.” Circulation vol. 137,3 (2018): 222-232. doi:10.1161/CIRCULATIONAHA.117.028021
  28. Andrew, P J, and B Mayer. “Enzymatic function of nitric oxide synthases.” Cardiovascular research vol. 43,3 (1999): 521-31. doi:10.1016/s0008-6363(99)00115-7
  29. Karim, M et al. “Effects of cocoa extracts on endothelium-dependent relaxation.” The Journal of nutrition vol. 130,8S Suppl (2000): 2105S-8S. doi:10.1093/jn/130.8.2105S
  30. Desideri, Giovambattista et al. “Benefits in cognitive function, blood pressure, and insulin resistance through cocoa flavanol consumption in elderly subjects with mild cognitive impairment: the Cocoa, Cognition, and Aging (CoCoA) study.” Hypertension (Dallas, Tex. : 1979) vol. 60,3 (2012): 794-801. doi:10.1161/HYPERTENSIONAHA.112.193060
  31. Ried, Karin et al. “Effect of cocoa on blood pressure.” The Cochrane database of systematic reviews vol. 4,4 CD008893. 25 Apr. 2017, doi:10.1002/14651858.CD008893.pub3
  32. Wan, Y et al. “Effects of cocoa powder and dark chocolate on LDL oxidative susceptibility and prostaglandin concentrations in humans.” The American journal of clinical nutrition vol. 74,5 (2001): 596-602. doi:10.1093/ajcn/74.5.596
  33. Parthasarathy, S et al. “Implications of lag time concept in the oxidation of LDL.” Free radical research vol. 28,6 (1998): 583-91. doi:10.3109/10715769809065814
  34. Fuhrman, B et al. “Consumption of red wine with meals reduces the susceptibility of human plasma and low-density lipoprotein to lipid peroxidation.” The American journal of clinical nutrition vol. 61,3 (1995): 549-54. doi:10.1093/ajcn/61.3.549
  35. Baba, Seigo et al. “Continuous intake of polyphenolic compounds containing cocoa powder reduces LDL oxidative susceptibility and has beneficial effects on plasma HDL-cholesterol concentrations in humans.” The American journal of clinical nutrition vol. 85,3 (2007): 709-17. doi:10.1093/ajcn/85.3.709
  36. Grosso, Giuseppe et al. “Dietary Flavonoid and Lignan Intake and Mortality in Prospective Cohort Studies: Systematic Review and Dose-Response Meta-Analysis.” American journal of epidemiology vol. 185,12 (2017): 1304-1316. doi:10.1093/aje/kww207
  37. Ivey, Kerry L et al. “Flavonoid intake and all-cause mortality.” The American journal of clinical nutrition vol. 101,5 (2015): 1012-20. doi:10.3945/ajcn.113.073106
  38. Ponzo, Valentina et al. “Dietary flavonoid intake and cardiovascular risk: a population-based cohort study.” Journal of translational medicine vol. 13 218. 8 Jul. 2015, doi:10.1186/s12967-015-0573-2
  39. Zamora-Ros, Raul et al. “Dietary flavonoid and lignan intake and mortality in a Spanish cohort.” Epidemiology (Cambridge, Mass.) vol. 24,5 (2013): 726-33. doi:10.1097/EDE.0b013e31829d5902
  40. Greenberg, James A et al. “Chocolate intake and heart disease and stroke in the Women’s Health Initiative: a prospective analysis.” The American journal of clinical nutrition vol. 108,1 (2018): 41-48. doi:10.1093/ajcn/nqy073
  41. Ho, Yuk-Lam et al. “Chocolate consumption and risk of coronary artery disease: the Million Veteran Program.” The American Journal of Clinical Nutrition, Volume 113, Issue 5, May 2021, Pages 1137–1144, https://doi.org/10.1093/ajcn/nqaa427