A Critical Analysis of Eggs and Cardiovascular Disease (The Eggscapades)

Matthew Madore

Matthew Madore

Author at MyNutritionScience

Author Profile

Background

Over the past few decades, one of the most commonly discussed and volatile topics in the field of nutritional science has been the relationship between egg consumption and chronic disease risk. Of the chronic diseases eggs have been implicated in, the most frequently recognized and discussed is cardiovascular disease (CVD, which is mainly a composite outcome including stroke, myocardial infarction, and heart failure), largely due to their saturated fat and cholesterol content. It has been suggested that because saturated fat and cholesterol can increase serum low-density lipoprotein (LDL) cholesterol and apolipoprotein B (apoB), which have been deemed causal in atherosclerotic cardiovascular disease (ASCVD) through decades of exhaustive research1, higher egg consumption could increase CVD risk. 

Despite this relationship providing a seemingly valid reason to believe eggs can increase the risk of CVD, results from studies on the relationship between the two have been incongruous to say the least. Consequently, every few years, if not months, the public is subject to a seemingly infinite cycle of news reports; one will claim that new research suggests eggs increase CVD risk and then shortly thereafter another will claim they don’t impact (or even decrease) risk. Like many other nutrition and health related topics, this instills widespread confusion and can leave people uncertain of who and what to trust. They may even throw up their hands in confusion and forgo any efforts to learn about and apply valuable dietary or lifestyle modifications conducive to better health. These sort of reactions are incredibly frustrating and disheartening for health and nutrition professionals who are truly doing their best to educate the lay public about the role of nutrition and other lifestyle habits in achieving and maintaining long term health. In an effort to address some of this confusion and help prevent subsequent frustration, this article will review the results of randomized controlled trials, a few recent meta-analyses of prospective cohorts, and food substitution studies pertaining to egg consumption and the risk of CVD. Hopefully the discussion will offer insight into why results have been conflicting, what the totality of the evidence suggests about this relationship, and important factors to consider when analyzing the effect of a particular variable on an outcome of interest. Further, ideally, the implications of its content will extend well beyond this particular topic and to many others besieged by similar contradictory results, confusion, and frustration. 

The Impact of Eggs on LDL/ApoB

Considering that the hypothesis eggs may increase the risk of CVD emerges from an increase in serum LDL/apoB as a result of their cholesterol and saturated fat content, their effect on lipids has been studied extensively in randomized controlled trials throughout the past couple of decades. Although both saturated fat and cholesterol have repeatedly demonstrated an ability to increase LDL and apoB (with cholesterol possessing a non-linear effect that will be touched on later)2-4, it is always best to confirm a specific effect can be attributed to the food/s in question. Foods are more than singular nutrients and comprised of varying proportions of fatty acids, nutrients, and compounds that may have differential effects on lipids. That said, a recent meta-analysis of over 60 randomized controlled trials did exactly this. It revealed fairly unequivocally that interventions increasing egg intake to over one per day significantly increase serum LDL and apoB in fewer than 12 weeks, an effect strongest in higher quality crossover trials and with higher doses5. The increases in LDL and apoB seen when pooling all trials were 7.3 mg/dL and 5.8 mg/dL, respectively. These are fairly small changes, however it should be kept in mind they are due to intake of only a single egg per day and are certainly not negligible when it comes to CVD risk. Furthermore, it’s important to note that these significant increases were seen even with the inclusion of at least a dozen trials containing characteristics that could have biased the effect towards null; such as energy restriction in the intervention group and high saturated fat and/or cholesterol comparator diets. The former is important since weight loss itself can elicit significant improvements in lipids6, and the latter because if the baseline or comparator diet have similar dietary cholesterol or saturated fat intakes, the difference when comparing to the egg intervention group would be small or absent depending on the presence or lack of contrast in nutrient intakes. Also worth noting is there remains uncertainty as to whether or not baseline cholesterol intake influences the impact of dietary cholesterol on LDL, as earlier work suggested such, but it was not seen in the latest meta-regression (see below). This may also be important since higher baseline cholesterol intakes would blunt the effect seen with adding more if it were true. Per these considerations, when focusing on trials that compared eggs against egg substitutes, the effect on both LDL and apoB were more notable (9.3 mg/dL and 6 mg/dL), clearly emphasizing the importance of having a lower cholesterol/saturated fat comparator. 

Therefore, in light of the fact that the most recent and comprehensive meta-analysis of randomized controlled trials revealed the consumption of one egg or more per day for under 12 weeks significantly increases serum concentration of causal intermediates for CVD (LDL, and more importantly apoB), it would be rational to infer that greater egg consumption would increase the risk of CVD. Curiously, while there are large prospective cohorts indicating this is true, the two newest meta-analyses of prospective cohorts (Godos et al. 2020 and Krittanawong et al. 2021) suggested that “up to one egg per day” decreased CVD risk significantly and that one or more eggs per day had no significant impact on CVD events but decreased coronary artery disease (CAD/CHD) risk, respectively7,8. Since it’s not feasible to carry out RCTs observing the impact of egg consumption on these outcomes, the next highest tier of evidence would be these meta-analyses of prospective cohorts suggesting that consuming one or more eggs per day does not significantly increase CVD events. As such, many people might quite reasonably assume that this is the true effect without further question, but to do so would ignore a huge number of important considerations that underpin the validity of the meta-analyses’ results. To underscore why these considerations are crucial when interpreting the results of meta-analyses, and why there is discordance between RCT and prospective cohort findings, the following sections include a breakdown of all the cohorts included in these two recent meta-analyses. These will serve to identify and discuss said factors and how they can confound results. 

Krittanawong et al. Meta-Analysis 

As mentioned, the two most recent meta-analyses observing the influence of egg consumption on CVD risk are those of Godos et al. and Krittanawong et al. Given that Krittanawong et al.’s meta-analysis was published earlier, it will be covered in this discussion first. Their analysis included 23 prospective cohorts and focused on 3 primary outcomes; a composite of CVD, CHD, and stroke. Correspondingly, they carried out a meta-analytic summation for each of these and reported the changes in risk when comparing “no eggs or one egg per day” to “more than one egg per day”, which produced the results seen below. 

The lack of a definitive comparator here is a major problem. If the change in CVD risk is nonlinear, then the change from zero eggs to more than one egg would be of a different magnitude than the change from one egg to more than one per day. Such a difference could dilute any effect on CVD risk from  increasing egg consumption from a baseline of zero, and therefore result in a biased estimate. Even if this was the only issue, it undoubtedly needs to be addressed since it hampers their analysis. Unfortunately, this was far from the only problem. While the analysis performed for CVD imported the HRs for each individual outcome (coronary artery disease, stroke, and heart failure), the following section of this article discusses only the individual cohorts focused on composite CVD outcomes for the sake of brevity.

Prospective Urban Rural Epidemiology Study9

In this study, Deghan et al. used data from three large prospective cohorts, PURE (the Prospective Urban Rural Epidemiology study), ONTARGET, and TRANSCEND to investigate what effect increased egg consumption had on cardiovascular incidence and mortality, among both healthy individuals and those with pre-existing cardiovascular disease. Additionally, they also chose to observe what effects egg consumption had on blood lipids. Their results suggested that overall a higher egg intake (one or more eggs per day) was not associated with a change in the risk of the composite CVD outcome. Individually, the results from PURE indicated that the same intake might decrease myocardial infarction risk, whereas those from ONTARGET and TRANSCEND showed no clear association. Only the results from PURE were included in the meta-analyses, so they will be the main focus here. 

While the total sample was substantial and the PURE cohort avoided a common issue among many of the other cohorts, a small intake range, these are essentially the only positive attributes of the analysis. First, the use of the PURE cohort as a major data set is concerning when taking into account the multitude of inherent issues it has, as discussed in various responses to the original publication utilizing the same population10-14. In these responses, it is thoroughly outlined that lower saturated fat/cholesterol (and subsequently higher carbohydrate) intake groups were mostly malnourished populations living primarily on refined carbohydrates, which are known to contribute negatively to cardiovascular health. Given this observation, it would not be difficult to compare even higher risk populations to that same group as a referent and show little or no difference in cardiovascular outcomes/overall mortality. As reflected in the original PURE publication’s supplement, the lowest quintile of saturated fat intake has a high incidence of cardiovascular/all cause mortality, cardiovascular disease composite incidence, and MI. This demonstrates that its use as a comparator is very problematic, since it would be difficult for outcomes in the higher intake groups to be much worse. Unsurprisingly, there appears to be a decrease in risk when going from low to high intakes. However, it would be inappropriate to conclude that increasing saturated fat/cholesterol intake is harmless or beneficial because there is a slight decrease in risk compared to a group with an already high incidence (due to impoverishment and poor diet quality). The same reasoning would apply to egg intake, and this is outlined by the presence of a similar trend across egg intake quantiles in Dehghan et al.’s analysis.       

Furthermore, China’s nutrient data (especially fat intake) was also shown to be off by a large margin when compared to data from national surveys during the same period, signalling major dietary records issues within at least one of the main countries included in the PURE cohort. Without getting into too much detail about additional issues, the aforementioned replies that various researchers have made to the original publication point out not only these flaws, but many more. It’s quite disappointing that this data set keeps being used despite such a large amount of errors, and it’s susceptibility to confounding by multiple factors. 

These details aside, the finding of “no significant difference” in blood lipids across quintiles of egg intake is expected and pretty invaluable given the recent meta-analysis of over 60 RCTs demonstrating that >1 per day significantly increased apoB and LDL. LDL was elevated across all quintiles, and  apoB just barely missed significance for an upward trend.

Countries with higher egg intakes also happened to have lower red meat and slightly higher fruit/vegetable intakes (these were adjusted for but it’s worth noting background diet quality trends upward). Other food and nutrient intake data stratified by egg intake would be extremely valuable here to assess whether other major differences in dietary patterns could have influenced lipids, and therefore MI risk, to a greater extent than the minor changes in egg intake. 

Lastly, if you look at the PURE subgroups stratified by region, almost all of them show the lowest risk of the composite outcome (CVD) is in the second and third lowest intake groups, with many being significantly lower than the already high referent group. Mortality also trends upward and becomes significant in the second-highest egg consumption group for PURE when excluding those with no established disease at baseline, then drops back to a null point estimate in the highest with wide CIs.

Although complete dismissal of results from the PURE cohort would be ill-advised, including them in a meta-analysis is highly problematic for the numerous reasons outlined here and would likely introduce substantial heterogeneity.   

Million Veteran Program15

Djousse et al.’s publication is an excellent demonstration of why considering the amount and range of egg intakes, pre-existing disease, and appropriate adjustments (especially avoiding the inclusion of causal intermediates) are incredibly important. Unfortunately, it will shortly become clear that this makes their research fall into the minority. For this study, they used data from the Million Veteran Program, a large prospective cohort of over 700,000 US veterans. A little over 100,000 US veterans free of cardiovascular disease were included in their analysis to ascertain the effect of egg consumption on the risk of having an MI. Despite a fairly short mean follow up of 3.24 years, there were an impressive 10,000+ new MI cases. When considering a range of intakes spanning from less than one egg per month to 2 eggs per day, there was a significant trend towards increased risk of myocardial infarction with increased intake. Compared to one egg per month, groups consuming 5 or more eggs per week had a significantly higher risk, unsurprisingly strongest in overweight individuals. 

Although there is certainly room for improvement, such as a longer follow up and slightly more varied demographic, this study was of good quality and more capable of providing valid results than many others. 

That being said, the hazard ratio values imported into Krittanawong et al.’s meta-analysis do not reflect those of the more than 1 egg per day intake group, but instead appear to be consistent with the 1-3 eggs per month group, which had an HR of 0.93 (0.85, 1.02) instead of the 1.13 (1.00, 1.28) seen in the highest intake group. This is a massive error that needs to be corrected for the forest plot to accurately reflect a comparison of over 1 egg per day to none.   

China Kadoorie Biobank16

In this publication, Qin et al. focused on results from the China Kadoorie Biobank cohort. This cohort definitely had potential to provide high quality evidence and important insight on the relationship between higher egg consumption and the risk of CVD. It provided data from 84,000 CVD cases over a median of 9 years in a sample of over 460000 Chinese adults aged 30-79. Despite these strengths, differences in sample population characteristics (both within the sample itself and compared to other cohort’s samples) and the nature of their egg consumption habits prohibited the analysis from effectively discerning any appreciable changes in CVD risk from increased egg consumption. Shockingly, the authors offered an over confident conclusion that demonstrated either nescience or dismissal of the collection of issues that should have strongly discouraged them from making such a strong closing statement. 

The bold comment made at the end of their publication asserted: “Our findings suggested that daily egg consumption (<1 egg) was associated with lower risk of CVD, IHD, MCE, haemorrhagic stroke and ischaemic stroke among Chinese middle-aged adults.” The basis of this statement was that, in their analysis, greater egg consumption (of a measly average of  0.76 eggs per day) compared to rare consumption (an average of 0.29 per day) resulted in a significant reduction in outcome risks. 

Not only was the difference between the highest and lowest group less than half of an egg a day, but both groups had incredibly low intakes overall. Someone concerned about potential increases in CVD with greater egg consumption would find little value in group comparisons for consuming under one egg to consuming slightly less than that. Such data would only be pertinent to the risk attributed to this miniscule intake difference. In reality, it would be extraordinarily naive to believe that any differences in outcomes observed across groups were attributed to the addition of less than half an egg, especially considering there were a multitude of far larger differences between other variables across said groups, that despite extensive adjustment would be hard to fully eliminate as confounders. 

Of these disparities, one of the most shocking was that relevant to the socioeconomic status of high and low egg consumers. About 48% of the high intake group had a middle or high school education, and 8% had a college education or greater, compared to under 40% and 3% in the low intake group. Also, about 50% of the high group had an income of over 20,000 yuan per year, and about 30% had 10,000-19,999, whereas the low group’s distribution was 35% and 30%, respectively. Lastly, almost 60% of the high group lived in an urban residence while only 30% of the low group did, only further emphasizing the socioeconomic gap. Unsurprisingly, subgroup analysis revealed significant interactions between some of these factors and many of the cardiovascular outcomes, for which the high intake group’s distribution was favorable in many instances (especially area and education level). On top of these issues, no data on lipids or detailed nutrient intake were provided, which are incredibly important to consider in this type of research. 

Henceforth, while this cohort might have appeared promising with respect to its ability to delineate the relationship between egg consumption and CVD risk due to its size and follow up, it was characterized by flaws that rendered it incapable of doing so. Pooling it with other publications in a meta-analysis only comparing low to high intakes would be inappropriate. 

Guangzhou Biobank Cohort Study17

Xu et al.’s investigation of greater egg consumption and the risk of dying from CVD used results from the Guangzhou Biobank Cohort Study, which similar to the Kadoorie Biobank, had the groundwork for a robust analysis by including a large sample size of around 28000 older adults with an average of 10 years follow-up. Their results indicated that higher egg consumption did not significantly affect the risk of CVD or any other outcomes considered. 

Sadly, major differences (and even a lack thereof) between the high intake and low intake comparators inhibited its ability to truly ascertain the effect of higher egg intake on CVDrisk. Like many other cohorts, although the factors with glaring differences across quantiles were adjusted for, their large magnitude makes it hard or near impossible to prevent confounding effects. Consequently, said adjustments would have reduced the already relatively small sample in the “high” group even more. 

The “high” group, consuming about one egg a day (vs ~1 a week) had a much higher proportion of the sample that was physically active and had a higher income, education, and fruit, vegetable, nuts, dairy and fiber intakes, whereas differences in saturated fat, and most remarkably cholesterol, were miniscule. The change in cholesterol from the low to high group was only about 20-30 mg, and the highest intake group didn’t even consume 200 mg per day even though the addition of a single egg should at the very least add about 200 mg, meaning that the lower groups were making up for the lack of cholesterol from eggs with cholesterol from other foods, and the group supposedly consuming one per day were likely not at all. This alone virtually eliminates the possibility of evaluating the effect of egg consumption on CVD risk since the cholesterol content (and to a lesser extent saturated fat) of eggs is the main characteristic that instills concern for CVD risk. With a difference equal to almost one tenth of a single egg’s cholesterol content, and no more than a single gram difference in saturated fat per day, it should be obvious that there is no way a comparison between these groups is appropriate. Or, if any miniscule effect does exist then it would be overcome by far larger differences in a host of other important dietary and lifestyle habits. 

The observation that higher egg consumption groups had significantly lower LDL, triglycerides, and blood pressure should have been expected given the higher activity level, the  ~1% difference in saturated fat intake (maximum), the tiny difference in cholesterol intake, and the much larger differences between fruit, vegetable, nuts, and dairy intake (all conducive to better lipid profiles) when going from “low” to “high” egg consumption. 

Interestingly, the authors somehow decided this might indicate “beneficial” effects of egg consumption. This is astounding given the aforementioned differences and how this would influence these metrics, as well-demonstrated in numerous RCTs. All in all, this is, unfortunately, another cohort where it would be egregious to assert the results are anything but a demonstration of other dietary and lifestyle effects that have repeatedly shown to impact CVD risk beneficially.   

Caerphilly Prospective Cohort Study18

This analysis of the effect of egg consumption on CVD risk and related biomarkers included data from two separate cohorts of healthy men and women in the UK. However, only the Caerphilly prospective cohort study (CAPS) was included for assessing changes in CVD risk according to egg consumption. CAPS included around 1800 men aged 45-59 who were followed for a mean of just under 23 years. While the follow-up length was very long and no problematic adjustments were made, the cohort certainly had a few notable characteristics that would undoubtedly impact their findings. 

First, like quite a few other cohorts included in the meta-analysis, the range in egg intakes was small. The lowest quintile consumed less than one per week, and the highest only more than 5 per week, equating toa difference of less than a single egg per day. Although this would strongly hinder the ability to pick up on changes in CVD risk, especially concomitant with the somewhat smaller sample of only a few thousand individuals (~14% of the total); the authors still observed a significant increase in stroke risk with higher (>5 per week) egg consumption. That being said, no differences in myocardial infarction and all cause mortality risk were observed, but looking over the assessment of dietary intake and biomarkers related to CVD risk, it’s very evident why this was the case. 

Intake of saturated fat was way beyond the typically recommended 10% of kcal for each group, the point beyond which participants are at higher risk to begin with. For the lipid biomarkers pertinent to myocardial infarction risk (i.e., LDL, triglycerides, HDL, etc.), values were virtually identical in all groups, with none of them being remotely close to ideal values and signifying increased risk across the board. 

This is reflected in the frighteningly high CVD event rate within all groups, with none of the numbers dipping below 35% of the sample, while half or more died from any cause. 

Considering this, even in the case that a difference in lipids from lowest to highest consumption groups was facilitated by higher egg intake, the fact that the low group (the comparator for their risk assessment) was already at an increased risk in comparison to those with normal values, the risk differences would be understated at best. That is, if there was enough power to detect it. One last observation worth noting was that even with the slight difference in egg intakes, there was a significant trend of increased blood glucose and Hba1c at multiple follow up times, which is certainly not a desirable outcome.

Alas, despite being well-designed and including a longer follow up, we should exercise caution in interpreting the results and reasoning their inclusion in meta-analyses of low-high intakes due to the sample size, span of the population’s egg intake, other dietary factors, and shared poor biomarker values 

Physician’s Health Study I19

Prior to their 2019 analysis of the Million Veteran Program, Djousse et al. used data from the Physicians’ Health Study I (PHS I) cohort to investigate if egg consumption correlated with CVD incidence and total mortality. The PHS I cohort contained around 21,000 US male physicians aged 40 to 85 with an average follow up of 20 years, during which over 1500 myocardial infarctions (CAD/CHD), 1300 strokes, and 5100 deaths were recorded. Subjects were split into five groups based upon their weekly consumption of eggs, ranging from less than one per week to 7+. This is a somewhat appropriate range of egg consumption. However, it is unfortunate that exact mean/median intakes were not recorded instead of only rough categorical distributions within this range. Furthermore, only around 8% of the sample consumed over 7 eggs per week, but the longer follow up prevented there from being too small a number of CHD cases in this group. The results are still puzzling regardless.

While there was no significant difference in the risk of CHD or stroke in the highest egg consumption group compared to the lowest, there was a significant increase in all-cause mortality present in both individuals with and without diabetes, strongest in the former. 

Of particular concern was that the total mortality cases equated to just under 40% of this entire group, with none of the others surpassing 25%. This raises questions as to what these individuals were dying from at such a higher rate if not CHD/stroke (two of the most common causes of death), and if it could have influenced the ability to detect an increase in one or both of these diseases. Another concern is the lack of more detailed nutrient/food intake. The small amount available seems to hint at the notion eggs typically replace “breakfast cereal”, which by today’s standards likely means high sugar and refined carbohydrate cereals, both of which are detrimental to cardiovascular and overall health.     

Nonetheless, this cohort was better than many of the others as a whole, despite a few major questions that remain unanswered with the presented data, the answers to which could definitely have a non-negligible impact on the relationship between higher egg consumption and CVD risk. Finally, their hazard ratio for myocardial infarction was included twice in the meta-analysis under two different years, despite clearly being the same data set as indicated by the identical point values and confidence intervals. Hopefully it doesn’t need to be said, but this is a major problem. 

First National Health and Nutrition Examination Survey20 

Using data from the First National Health and Nutrition Examination Survey (NHANES I), Qureshi et al. carried out an analysis in order to uncover the effect of egg consumption on stroke,  CHD/CAD, and mortality risk. The first NHANES followed a sample of around 9700 men and women aged 25-74 in the United States for a mean of 16 years, during which over 650 strokes and 1580 myocardial infarctions (CHD/CAD events) occurred. Subjects were split into 3 groups, those who consumed less than 1 egg per week, 1 to 6 eggs per week, and more than 6 per week (~1 per day). After multivariate adjustment, there was no significant increase in CVD risk when comparing the highest tertile of egg intake to the lowest overall, but the story doesn’t end there.

Viewing the results, it is evident that higher egg consumption appeared to be associated with higher CHD and mortality in the age-adjusted model. However, the adjustment for serum cholesterol attenuated these associations expectedly, considering it is the major causal intermediate by which eggs are expected to impact CVD risk. Risk was still significantly increased in those with diabetes despite this adjustment, for which the bottom of the risk ratio’s confidence interval sat very close to the null but extended all the way up to almost 4. 

Given the inclusion of serum cholesterol in the multivariate model it would be outright senseless to import the results of this analysis into a meta-analysis due to the potential to bias any effect towards null. Inclusion of cohorts that do this is such a frequent yet completely avoidable issue in meta-analyses pertaining to the risk of dietary habits on CVD risk.       

Health Professionals Follow Up Study and Nurses Health Study21 

In this analysis of the Health Professionals Follow Up and Nurses Health Studies, Hu et al. set out to determine the effect of egg intake on CVD incidence. These two cohorts included about 38,000 men aged 40 to 75 and 80,000 women aged 34 to 59, who were followed up for about 8 and 14 years, respectively. Results suggested that more than or equal to one egg per day did not increase the risk of CHD or stroke incidence compared to “<1 per week”, although there was a significant increase in CHD risk among those with diabetes. 

While nothing stands out as especially problematic (though the number of subjects in the higher intake groups may have been a limiting factor since they contained under 10% of the samples), the authors continue on to discuss a collection of reasons they suspect may have prevented them from observing a change in risk for the total sample. Initiating this discussion, they note that previous controlled feeding trials have indeed confirmed that dietary cholesterol increases serum cholesterol. Looking back to the result of these studies, they mention that an increase of one egg per day on top of the baseline cholesterol intake among the lower consumption group (~200 mg) would only be expected to increase serum cholesterol 4%, and the risk of CHD by 8%. They then remark that their analysis would likely not even be capable of picking up on such a change in risk (especially given the smaller sizes of the groups with the highest intakes) based on these considerations. Furthermore, they also continue on to comment that even the lowest dietary cholesterol intakes of individuals in these cohorts may already influence CHD risk compared to an individual consuming none, and that due to the potentially hyperbolic nature of the effect of dietary cholesterol on serum cholesterol that any minor increase in risk from a baseline already well above zero could be even more difficult to ascertain. Unfortunately, they did not have large enough samples of people with very low intakes to make this comparison or analyze higher intakes (above one egg per day). Finally, one other important comment they made was the following: “In most egg feeding studies, intakes of other nutrients such as fatty acids, carbohydrates, and protein were balanced between egg and no egg groups so that only dietary cholesterol varied. In our cohorts, participants who consumed more eggs had lower intakes of carbohydrates, suggesting that, in reality, people often substitute eggs for carbohydrate-rich foods such as breakfast cereals.” The suggestion that eggs may replace breakfast cereals is indeed supported by the nutrient intake data, revealing that over double the amount of individuals in the low egg group were consuming cereal almost daily compared to those in the high group. 

Given that replacement of saturated fat (which has a far stronger influence on serum lipids) with refined carbohydrates (such as those present in a vast majority of breakfast cereals) does not reduce or may increase CHD risk22,23, this is an important detail to take into consideration. In a case such as that observed in this cohort where subjects are replacing processed cereal with eggs, there will likely not be a propensity towards increased risk, but if both are harmful to cardiovascular health, swapping one with the other and observing no change in CVD risk would not be very informative or valuable. 

So overall, although these cohorts were composed of two large samples that encompassed a fairly large amount of cardiovascular events, the authors were very forward in revealing that multiple factors were capable of causing their analysis to turn up null results. Further, there was an increase in risk seen in individuals with diabetes, which is certainly worth considering since rates of diabetes have surged within the past few decades.    

Lifetime Risk Pooling Project24

In what is very likely the highest quality study included in the meta analysis, Zhong et al. set out to observe the relationship between eggs/dietary cholesterol intake and incident cardiovascular disease risk using a sample of just under 30,000 adults without CVD from 6 prospective cohorts in the United States that were part of the Lifetime Risk Pooling Project. Subjects were followed up for a mean of 17.5 years, multiple dietary assessments were conducted, and there was a broad range of egg intakes (near 0 per day to 2.5 per day).

Their results suggested that both an increase in 300 mg dietary cholesterol and only a half an egg a day were associated with a significant increase in CVD and all cause mortality, both of which persisted in their extensive sensitivity analyses. 

Furthermore, adjustment for egg or dietary cholesterol intake attenuated the significant associations, strongly indicating they drove the increase in risks observed. Despite Zhong et al. very clearly stating that they reported the model excluding adjustments for causal intermediates (especially serum lipids) because they could (and did) nullify increases in risks due to higher cholesterol/egg consumption, the authors of the meta analysis including this study chose the single model that adjusted for lipids and dietary cholesterol intake. If one were to be generous, this is unbelievably naive, but it’s hard to imagine such naivety among experienced researchers, suggesting it may even be purposeful dishonesty.

Though this analysis was easily the most capable of providing the quality data needed to answer the main question of the meta-analysis, the authors did not even allow for it to contribute relevant risk estimates. The values imported were those from the single explorative model that intentionally adjusted for the main risk mediators and this is, to put it frankly, unacceptable. It certainly requires correcting.  

Korean Genome and Epidemiology Study25

The KGES cohort used by Jang et al. in their analysis included just under 10,000 healthy Korean adults followed up for an average of 7 years, during which 570 cases of CVD occurred. The sample encompassed in this cohort is almost an ideal population for assessing the influence of egg intake on cardiovascular disease risk given the low mean baseline serum lipids and decent proportion of the population with close to zero egg intake. Unfortunately, the range in intakes was remarkably small, with the lowest quartile consuming ~0.1 eggs per week and the highest only 4 per week, alongside the fact fruit and vegetable intakes, education, and income also increased notably from the low egg intake group to the high group. 

As would be expected, no difference in the risk of CVD was observed when comparing the highest to the lowest intake. However, to put this into perspective, 4 eggs provide under 300 kcal (280) calories , and 800 mg cholesterol, and the high intake group was consuming this many per week. Furthermore, this lends to an even smaller difference of ~3 eggs per week compared to the low intake group, which would be the difference considered with respect to the calculations for CVD risk. 

Funnily enough, even considering these issues, subgroup analyses revealed that those with type 2 diabetes at baseline (a group at a much higher risk of experiencing a cardiovascular event in a shorter follow up, and therefore offering potential to observe differences that might only manifest over longer follow ups otherwise) consuming more eggs had a significantly higher risk of incident CVD.  

Given the lack of information about saturated fat and dietary cholesterol intakes among groups, as well as the extremely small range of intakes considered, it would be almost preposterous to think that any appreciable changes in risk resulting from higher egg intake could be gleaned from a cohort like this. The sample would need to be enormous and followed for decades to gain valuable information. 

Prevención con Dieta Mediterránea Study26  

This publication was a secondary analysis of the PREDIMED cohort by Diez-Espino et al. that intended to determine the impact of egg consumption on cardiovascular outcomes. When looking over baseline egg intakes, they did not find a significantly higher risk of incident CVD in the group with the highest egg consumption. 

However, the highest intake group consumed “>4 eggs per week” and was compared to “<2 eggs per week”, as such there was hardly even a difference in intake across tertiles. Furthermore, the low egg group had dietary cholesterol intakes of about 300, whereas the highest group had intakes of about 500. 

This is problematic because the low group’s baseline intake was well above what would be expected to impact lipids significantly. Given that the slope representing the change in lipids with increased cholesterol consumption may be steepest when increasing from 0 to 300-400 mg dietary cholesterol, the change that would be observed from 300-500 could be appreciably smaller. Any risk difference between these numbers would be much more difficult to pick up in statistical analyses. This issue would only be compounded by the incredibly small sample (214, or 3% of the sample, in the >4 eggs per week group compared to 4493 and 2509 in the <2 eggs and 2-4 eggs per week groups, respectively) in the “high” intake group. Lastly, the fact that the PREDIMED trial involved the assignment of many participants to a Mediterranean-style diet lower in saturated fat and purposely intended to reduce CVD risk, an instance where the benefit would far surpass any effect that egg consumption may exert, raises many concerns.   

Third National Health and Nutrition Examination Survey27

Using data from the Third National Health and Nutrition Examination Survey (NHANES III), Scrafford et al. conducted an analysis concerning the effect of egg intake on CHD and stroke mortality. The NHANES III cohort included just under 7000 men and slightly over 8000 women over 17 years of age, who were followed up for a mean of about 9 years. After being split into three groups based on weekly egg intake (less than one per week, equal to or more than 1 and less than 7 per week, or over 7 per week), there was no clear relationship between egg intake and the risk of CHD, and a decreased risk of stroke. 

Unfortunately, numerous major problems plagued this analysis. The first thing that stands out is the mean ages of the groups (ranging from 40-43) and the corresponding follow up times (~9 years, as mentioned previously). With such a large proportion of the sample not even reaching ages where CHD and stroke mortality typically begins to increase appreciably, this makes a lot of the population of little value for determining any relationship between egg consumption and CHD. This would limit the number of cases available for analysis, which is clearly evidenced by the total number of CHD events being below 400 and the wide confidence intervals of the hazard ratios. Also, oddly the third multivariable model included adjustments for saturated fat and cholesterol intake, the exact components of eggs that would be expected to affect CHD risk, and which would result in a bias towards the null. Finally, the dietary cholesterol intakes of the <1 egg per week groups were 288 mg and 203 mg for men and women, which as mentioned previously, make picking up an effect when increasing egg intake even more difficult. Likewise, the women’s cholesterol intake only increased ~140 mg in the high intake group and their saturated fat intake was actually lower, indicating they reduced intakes of these two nutrients from other foods as they increased egg intake, which again may impair the ability to pick up a change in risk and decrease the utility of this analysis in determining any effect that may exist.  

Seguimiento Universidad de Navarra Project28

In this study, Zazpe et al. set out to characterize the relationship between eggs and CVD using data from the SUN project, a Mediterranean cohort of about 14000 college graduates aged 38-41 who were followed up for 6 years. As mentioned while discussing the previous cohort, this is a curious choice of demographic to use to investigate the impact of dietary habits on CVD since overall incidence would likely be low and confined to those with extraordinarily bad habits. 

Expectedly, this is clearly reflected in the fact that less than 100 cases of CVD occurred, including less than 40 cases of myocardial infarctions and less than 30 of stroke. Furthermore, egg intake only ranged from <1 per week to >4 per week, which is not much of a difference and only compounds the aforementioned issue. 

All that being said, the authors closed off the results by remarking: “Finally, stratified analysis by median age (36 years) was not possible because of insufficient number of CVD events in the younger group. For this reason, we performed an additional analysis using a cutoff point of 56 years of age, which left a similar number of events in both groups (47 events among the younger participants and 44 events among the older participants). 

The HR for CVD for the highest egg consumption category compared with the lowest egg consumption category among participants with ⩽56 and >56 years were, respectively, HR: 0.47 (95% CI: 0.11–2.05) and 2.94 (95% CI: 0.93–9.30).” Not only does this emphasize the importance of the earlier comment regarding issues with using a younger sample, but it also seems to suggest that when considering an age group more likely to experience events during follow up, that risk probably does increase. However, given the incredibly small samples and high degree of uncertainty, using this analysis to draw any conclusions about the effect of egg intake on CVD risk would be an abysmal decision.  

Health, Aging and Body Composition Study29

Houston et al. used results from the HABC study to research the potential impact of eggs on CVD risk. HABC contained a cohort of just under 2000 community dwelling older adults aged 70-79, followed up for around 9 years. Results demonstrated that those consuming over 3 eggs experienced a significant increase in CVD risk, after extensive consideration and adjustment of other lifestyle/dietary habits. 

However, this risk was confined to (and very notable in) individuals with diabetes, which is not all that surprising. They would be at a much higher risk to begin with and more likely be impacted by small differences in intake through a shorter follow up period. Even non-diabetics appeared to lean towards a higher risk despite event rates being fairly high in each group. Furthermore, the observations in diabetics are consistent with previous findings that even minor differences in egg intake significantly impact risk of CVD. 

Although an older age group such as that included in this study is more likely to experience cardiovascular events at higher rates and therefore allow for a better powered analysis, only looking at risk according to dietary data in subjects’ later years without knowledge of prior dietary habits could be problematic, underscoring the value of long follow ups. This is because earlier and prolonged exposure to higher serum cholesterol has a far stronger influence on CVD risk that may far surpass any effect be attributed to differences in dietary patterns in the few later years of the participants’ lives encompassed in this cohort’s follow up. So while not perfect (though that is a far too idealistic goal), this cohort does seem to offer valuable data that would be much more appropriate to include into a meta-analysis than that produced by many of the others. 

Japan Public Health Center-based Prospective Study30

Following an earlier investigation using a different sample, in 2007, Nakamura et al. sought to observe the relationship between egg consumption, serum cholesterol, and coronary heart disease incidence using data from two Japanese cohorts under the Japan Public Health Center-based prospective study (JPHC). The two cohorts included about 20000 men and 21000 women, aged 40 to 59, and about 23000 men and 26000 women aged 40 to 69, who were followed up for 11 and 7 years, respectively. They split subjects into four groups by weekly egg consumption; <1 day per week, 1-2 days, 3-4 days, and “almost daily”. Using the almost daily group as the referent, they did not observe any significant differences in CHD incidence between any of the other groups, though leaving things off there would exclude a lot of important details.a

Similar to their publication focusing on NIPPON DATA80, they also reviewed data on serum cholesterol concentrations of a subset of patients, confirming that higher concentrations significantly increased the risk of CHD incidence. 

This observation provides part of an explanation as to why no differences in CHD incidence were likely seen across the different groups of egg consumption; since total cholesterol trended upwards with lower consumption, in addition to the fact that a significantly greater proportion of the low intake group was older, had higher blood pressure, cholesterol over 2200 mg/l (where CHD risk increased significantly), use of cholesterol-lowering drugs, lower fruit, vegetable, and fish consumption, and most importantly; more were intentionally restricting cholesterol intake. 

All of these taken together signify a non-negligible potential for the influence of reverse causality even in the presence of adjustments for these variables. In addition, the total range of egg intakes was not very large to begin with. There was no detailed nutrient data (were saturated fat, trans fat, and refined carbohydrate intakes higher or lower with increased/decreased egg intake?) and the overall number of CHD cases was not very high, as reflected in incidence rates that were far lower than other major countries (likely a product of the relatively lower age and short follow up of the sample). The culmination of all these factors indicates that parsing out significant effects on CHD rates from higher egg consumption would be extremely difficult, if possible at all, as any results would almost certainly be representative of other characteristics of the demographics unrelated to egg intake. 

Aborigines in the Kimberley Region of Western Australia31

Due to the paucity of data concerning health-related behaviors and coronary heart disease/all-cause mortality in Australian Aborigines, Burke et al. carried out an analysis on a sample of 256 female and 258 male Aborigines aged 15 to 88 from the Kimberley Region of Western Australia to determine behaviors that may influence these outcomes. After around 14 years of follow up, researchers determined that the consumption of more than just 8 eggs per month was associated with a significant increase in coronary heart disease incidence/mortality, even after adjustment for the main causal intermediate (cholesterol). 

However, although the authors mention adjusting for smoking, drinking, and physical activity in their statistical analysis section, these variables are not mentioned in the adjustments listed under the table listing hazard ratios for the outcomes. 

If they were truly not included in adjustments, this would be a huge issue. Even then, the very high proportion of men who smoked and/or drank would have substantially lessened the sample size and statistical power, which is definitely suggested by the wide confidence intervals. 

The small sample size, unclear adjustments, and cluster of unhealthy habits present in this sample should caution one against making any inferences from this data. Its inclusion in the meta-analysis is questionable despite agreement with data from RCTs observing the impact of eggs on atherogenic blood lipids. 

All in all, at best it may be arguable that two or three cohorts in this entire analysis were appropriately designed, contained relevant samples, encompassed a large enough range of egg intake, and did not have stark differences in sample characteristics across quantiles of intake that allowed for the assessment of increasing egg consumption on CVD risk. Also, multiple importation errors led to the inclusion of incorrect hazard ratios, a critical problem that requires attention. Henceforth, the null results their analysis produced were almost to be expected. That being said, the few studies that were of higher quality tended to show either a significant increase in CVD risk or lean in that direction with greater intake of eggs, consistent with their impact on LDL/apoB. Furthermore, the risk increase seen in individuals with diabetes was much more prevalent, even in many of the studies characterized by factors that one might think would almost guarantee a null result. So, while Krittanawong et al.’s meta-analysis falls short of addressing the question they set out to answer, it does help to highlight critical considerations for both the design of the included studies and meta-analyses as a whole. 

Godos et al. Meta-Analysis

With their meta-analysis, Godos et al. sought to elucidate if increasing egg intake affected the risk of CVD morbidity/mortality (including myocardial infarction/coronary heart disease (CHD), stroke, and heart failure), as well as of all three individually. In order to do so, they pooled together all the prospective cohorts they could find investigating the relationships between eggs and these outcomes. They also decided to conduct a dose-response analysis for these outcomes, which is an incredibly important consideration when carrying out research attempting to establish a connection between certain foods and the risk of various diseases (which should have been made fairly apparent in the discussion of Krittanawong et al.’s meta). While this decision would help reduce the potential of pooling together results from cohorts with substantially different intakes (and therefore also different changes in risk), it unfortunately leaves the analysis susceptible to many other issues previously covered. When all was said and done, the results of their analysis indicated that up to six eggs per week (which they considered one egg per day) was associated with a significantly reduced the risk of CVD, no effect on CHD or stroke, and a borderline significant increase the risk of heart failure (that became significant with over 6 eggs per week). However, despite the authors’ ambition and fairly solid methodology, their analyses included numerous cohorts with fundamental flaws similar to those seen in Krittanawong’s meta. Without accounting for these flaws, any ability they may have initially possessed to ascertain the true effect of egg consumption on CVD/CHD risk would be severely undermined. Accordingly, the following discussion will analyze each individual cohort included in their meta analyses for aggregate CVD (coronary heart disease and stroke) and coronary heart disease (CHD) incidence/mortality (the two outcomes studied in most detail), highlighting the aspects that are potentially problematic. Worth noting is that many of the cohorts included in the analysis were already covered in the discussion of the meta-analysis conducted by Krittanawong et al., so the following discourse will pertain only to those unique to this meta-analysis.

CVD Incidence/Mortality 

European Prospective Investigation into Cancer and Nutrition – Greece32 

Using a subset of just over 1000 diabetics from the EPIC-Greece arm, Trichopoulou et al. conducted an analysis exploring how increased intake of various foods impacted the risk of cardiovascular mortality. Their results demonstrated that among these diabetics, a one standard deviation increase in egg intake (10 g, or around ¼ of a large egg, from the mean sample intake of 10 g) was associated with a significantly increased risk of cardiovascular mortality. The HR for this increase was an impressive 1.54 (1.20 to 1.97), which remained significant after the exclusion of those who died in the first year of follow up. 

Unfortunately, the small sample had a very low incidence rate that would lead to lower statistical power (although this almost makes it more impressive that the analysis reached significance and didn’t have an extremely wide confidence interval). Since it was composed only of diabetics, generalized applicability would be limited. Lastly, follow up was not very long overall and the exposure contrast observed wasn’t very large, which injects a level of uncertainty into the findings, despite coinciding with those of other analyses. 

The Northern Manhattan Study33

Using data from a sub-cohort of about 1400 multi-ethnic subjects living in Northern Manhattan with a mean age of 66 and followed up for 11 years, Goldberg et al. attempted to characterize the association between egg intake and carotid intima thickness (cIMT) and clinical vascular events. In their analysis of continuous egg consumption per week, there did appear to be a significant influence of intake on combined cardiovascular events…until it was adjusted for LDL. Being the major causal intermediate responsible for potential changes in CVD risk with greater egg consumption, this adjustment alone is enough to completely wipe out the possibility of even picking up on an association between the two, but that’s not all. 

The range of the participants’ egg intakes was from less than twice a month to a whopping four times a week. Furthermore, the authors went on to comment: “In response to the frequency of whole egg consumption with a medium portion size of 2 eggs, 23% of the cohort reported that they consumed eggs less than once per month, 26% 1 time/month, 16% 2-3 times/month, 30% 1 times/week, 3% 2 times/week, 2% 3- 4 times/week, 1% 5-6 times/week, and nobody reported consuming eggs daily”.

This indicates that the group with the highest intake of eggs (if “high” is even appropriate here) was incredibly small, which would reduce statistical power substantially. Also, the tiny range of intakes would increase the chance that other factors could far outweigh any effect that could possibly be detected. Lastly, as mentioned by the authors themselves, there was a strong possibility for reverse confounding. Individuals at higher risk may have intentionally avoided eggs, in which case the low intake group would likely have a high event rate, and skew the risk ratios ascertained by comparing to the high intake group. The distribution of lipids, hypertension, triglycerides, and the use of lipid-lowering medication all appear to indicate it could very likely be the case. 

While authors reasoned a few differences might have counterbalanced the confounding (smoking, calorie intake, and lower physical activity), smoking rates were hardly different between groups, and the lower BMI and percentage of males in the high intake sample both suggest that the latter two reasonings aren’t feasible as higher caloric intake would be expected, and combined with the rate of intense physical activity it clearly wasn’t leading to greater accumulation of fat mass. Considering this array of issues, this cohort is definitely not very fit to answer the question posed by Godos et al.’s analysis.   

Cohort of Swedish Men and The Swedish Mammography Cohort34

In order to address a lack of studies covering specific outcomes in relation to egg consumption, Larsson et al chose to investigate how it affected the risk of heart failure, myocardial infarction, and individual stroke types. For their analysis, they used data from two Swedish cohorts, COSM and SMC, which were comprised of around 38000 men aged 45 to 79 and 33000 women aged 49 to 83 who were followed up for 13 years. Their results suggested that consumption of >1 egg daily had no effect on MI or stroke risk, but significantly increased the risk of heart failure in men. Worth noting is that only 3.7% of the entire cohort did not consume any eggs, and 2.7/2.1% of men/women had over one egg a day, which should certainly hinder the confidence in the findings within these groups. 

The authors mention a secondary analysis of even higher intakes and say that they don’t change the results, but unfortunately, these are not shown. Most importantly, looking over the data on nutrient intakes, there is no data on saturated fat, which prevents the consideration of whether or not total intake remained stable, increased, or decreased across quartiles. The latter case has the potential to substantially confound the results by offsetting the serum cholesterol raising effect of dietary cholesterol that would be expected to elicit any change in CVD risk. Furthermore, the lowest egg intake groups had dietary cholesterol intakes of 240 and 250 mg/day for men and women, whereas those in the highest groups had 480 and 450, respectively. Considering a large egg has around 200 mg of cholesterol, these numbers essentially reflect a comparison of 1 to 2 eggs worth of cholesterol per day. The fact that the lowest intake group already had an appreciable dietary cholesterol intake and that the LDL change in response to dietary cholesterol is hyperbolic suggests any effect of going from 200+ mg to 400+ mg will be very small. It would require a tremendous amount of statistical power to pick up on this, which is far from the case here given the small samples on both extremes of the spectrum. 

Additionally, the distribution of hyperlipidemia once again hints towards the possibility of reverse confounding, as it was by far most prevalent in the low egg intake group. Finally, around 20% of the low intake group experienced a cardiovascular event, so using this as a comparator and finding no increase in risk is still an unfavorable finding. In conclusion, though the cohorts included in Larsson et al.’s analysis were large, had a long follow up, and a high CVD incidence rate, the compilation of problematic characteristics rendered it much less likely to provide an accurate answer to the query at hand. 

Golestan Cohort Study35

Owing to the lack of studies on non-western countries investigating the relationship between various dietary protein sources and cause-specific mortality (all-cause, cardiovascular disease, and cancer mortality), Farvid et al. set out to characterize this relationship using the GCS cohort located in the Golestan province in Iran. This cohort included a sample of over 42000 disease-free men and women aged 36 to 85 that were followed up for 11 years. After adjusting for covariates, they determined that greater consumption of eggs had no significant effect on the CVD mortality risk, and may have even reduced all-cause mortality. 

With a large sample of healthy individuals, a relatively long follow up, and no adjustments for casual intermediates, this analysis initially appears to be of very high quality and has a solid framework for producing high-quality data. Unfortunately, it falls victim to one of the major issues so many of the other cohorts in this meta analysis do; the exposure contrast is remarkably small, not even representing a half of an egg per day when going from the lowest to the highest quintile of intake. It would be strange to think that this tiny of a difference would allow for meaningful determination of the change in risk associated with higher egg consumption, and there would be a good chance any effect that does exist would easily outweighed by other differences in dietary intakes (namely of saturated fats, trans fats, refined carbohydrates, etc.). Also, since the investigators’ main concern was with the effect of red meat on the outcomes considered, they only presented detailed information about the participant’s characteristics stratified by red meat intake, preventing the opportunity to further scrutinize other potential differences in dietary intake, socioeconomic status, and health status that may have occurred across quintiles of egg consumption. Therefore, at the absolute best these results suggest that very minor changes in egg intakes, from none to a median of just under half an egg a day, don’t meaningfully affect CHD risk in this demographic, which is not particularly valuable information. 

Spanish European Prospective into Cancer and Nutrition Study36

The cohort used in this analysis by Zamora-Ros et al. consisted of a large sample of 40,000 individuals aged 26-69 from 5 Spanish regions followed up for 18 years as part of the European Prospective Investigation into Cancer and Nutrition, hence the name “EPIC-Spain”. Once again, it is incredibly unfortunate that despite the potential to provide robust data that could characterize the association between egg consumption and cardiovascular disease risk, this cohort contained a collection of issues that rendered it unable to do so. Their main finding concerning CVD was that eggs did not significantly increase mortality risk. 

However, like many previous cohorts, the intake of eggs was not much overall and did not encompass a large range of intakes, with the highest and lowest quartiles being less than around 11/14g per day and greater than around 30/43g per day for women/men. The authors also note: “The percentage of non-egg consumers was 5.7% for women and 5.2% for men. The egg intake in this population was low to moderate; only 9.9% and 2.4% consumed ≥ 1 and ≥ 1.5 egg/day, respectively.”, showing how just a tiny portion of the sample even fell in the “extremes” (though the larger values are far from extreme) of low and high, leading to lower statistical power to detect meaningful differences even before adjustment. Additionally, there were vast disparities in hyperlipidemia prevalence at baseline when proceeding from low to high egg intake groups. In the low groups, 26% and 38% of the sample had hyperlipidemia compared to only 9% and 17% of the high intake groups. Also, the incidence of rates of myocardial infarction, stroke, hypertension, diabetes, and cancer was also significantly higher in the low groups. In addition to the higher mean age, all of these indicate they would be at a substantially greater risk of experiencing cardiovascular events/mortality, and suggest that a substantial proportion of the sample had “lower” intake only as a result of changing dietary habits to address these conditions. 

Even with extensive adjustment, it is a major problem that the highest intake group is pitted against a comparator expected to have much higher event/mortality rates from baseline as it would prevent, or at the very least strongly dampen the power of, risk assessments. These considerations combined with the low range of intakes and the much greater intake of multiple health promoting foods (olive oil, legumes, cereals, and to a lesser extent fruits and vegetables) in the high consumption group paint a clear picture as to why there would be no observed effect on CVD mortality, although this result is probably not trustworthy and likely not representative of the true effect.   

The Netherlands Cohort Study37

Using a nested-case cohort method on NCS, van deb Brandt et al. set out to determine the effects of different dietary protein sources on overall and cause-specific mortality (including CVD). NCS included two samples of men and women aged 55-69; around 120000 in the full cohort, and 5000 in the subcohort used for this analysis. Evaluating the dietary data and deaths in the main and subcohort (both followed up for ten years), their results suggested that those consuming the most eggs were not at a significantly higher risk of cardiovascular death. However, when taking a step back and observing other details, it becomes apparent that it was practically guaranteed this would be the outcome. Continuing the trend seen in the earlier cohorts, the difference between high and low egg intake here (the exposure contrast) was almost comically small. The mean intake ranged from 0 grams in the low group (which also had an extremely low number of person-years relative to other groups, around 8% of the total sample), to about 21 grams in the high group, or under half an egg daily. 

In addition to the analysis of the effect of egg consumption on CVD mortality being hindered by this minuscule range of intakes, and potentially by the small size of the comparator group, there was also no data on nutrient intake, health status, and socioeconomic status stratified by egg intake quartiles,. This prevents further consideration of how they varied among quantiles and the potential for reverse causality. Even just the small contrast in egg intakes among individuals in the sample should alert investigators that it would be short sighted to draw a conclusion regarding the impact of egg consumption on the outcomes of interest, not to mention the additional problems with the data set.  

Health Professionals Follow-Up and Nurses Health Study I/II38 

For the most part, Drouin-Charter et al.’s analysis of the results from three large US-based prospective cohorts, The Health Professionals’ Follow-Up Study and Nurses Health Study I and II, was well-conducted. These cohorts comprised a large sample and were followed for up to 32 years, suggesting they would likely be able to carry out high powered statistical analyses to accomplish the authors’ goal of assessing how egg intake might influence the risk of cardiovascular disease. Alas, a few characteristics could have caused minor issues, and one that was particularly troublesome. 

Looking over the sample characteristics collected at the midpoint of the cohort’s follow ups, a few things stand out. As discussed in one of the earlier analyses of the first two cohorts, eggs appeared to replace refined breakfast cereals fairly frequently. In this case, no change in risk would potentially just reflect going from one type of food detrimental for cardiovascular health to another. Aside from this, the most notable standout was that the proportion of the low egg intake group with hyperlipidemia was substantially higher than that of the greater intake group, and many more subjects also took statins in two of the cohorts (NHS I and HPFS). NHS II had a low overall number of individuals with hyperlipidemia and using statins, not all too surprising given the sample was much younger. The difference in hyperlipidemia prevalence between the high and the low groups in NHS I and HPFS were 11% (56% in the low consumption group) and 23% (54% in the low consumption group), respectively. Further, statin use was at least double for both of the low groups, 19% compared to 5% in NHS I, and 22% compared to 5% in HPFS. Considering the higher percentage of individuals in the low intake group with hyperlipidemia/taking statins, and the fairly steep downward trend in egg intake from the beginning of the follow up to the midpoint as depicted in supplementary figure 2, there is a high likelihood that the people in this group consumed such a low amount of eggs due to the fact they were at a higher risk of cardiovascular disease during the survey, and the consumption of dietary cholesterol and saturated fat are typically cautioned against for those at high CVD risk. As such, unless they were making drastic lifestyle changes and religiously adhering to medication protocols, the low intake groups would be expected to experience far more events, which was reflected in the cases/person years observed through follow up and makes the finding that there was no significant increase in risk with higher egg consumption fairly predictable.

While one might assume that adjustments could solve this issue, it is important to note that over half of the low intake groups had hyperlipidemia. These adjustments would decrease the already low statistical power of a group that was a small percentage of the total sample. Also, not only were the low intake comparator groups small and expected to have higher event rates according to baseline characteristics, but the high egg intake groups were even smaller. As quoted in the text: “In 1998-99, a total of 2524 of 203364 participants (1.24%) consumed at least one egg per day in the three cohorts. Of those, 414 of 203364 (0.20%) consumed at least two eggs per day”. Additionally, since NHS II had a much younger sample (women aged 30 to 55) and a shorter follow up, the event rates were low. Therefore, in the extremely small high intake (~1 egg per day) group they totalled only seven cases, which would severely impinge statistical analysis. This is clearly evidenced by the massive confidence intervals for this group’s hazard ratios, and only worsened in the sensitivity analysis looking at even higher (~2 eggs per day) intakes. 

So while it appears as if they included a decent range of egg intakes, it was at the expense of having very small sample sizes with low event rates that were virtually guaranteed to be underpowered, influenced by confounders, and hard to draw confident conclusions. In closing out the publication, the authors make note of the issue of low intakes, commenting that “However, mean egg consumption in the three US cohorts in our study and in cohorts included in the meta-analysis was relatively low. This consumption level should be taken into account when interpreting our results because most participants consumed one to less than five eggs per week, and relatively few participants consumed at least one egg per day.” Therefore, the best that could be extracted from this analysis is it demonstrates that when comparing a very small amount of individuals consuming one egg per day to those with the lowest consumption, of notably worse baseline health status, then the former is not at an increased risk of CVD. 

Prediction for Atherosclerotic Cardiovascular Disease Risk in China39

This publication covers the results of Xia et al.’s investigation into the effect of egg consumption on incident cardiovascular disease and all-cause mortality. Their analysis included over 102,000 adults free of cardiovascular disease from 3 cohorts located in various areas of China, allowing for an impressive 770,000+ person-years of follow up during which 4,848 cardiovascular events and 5,511 deaths occurred. Interestingly, their results suggested that there was a U shaped association with egg consumption and CVD/all cause mortality, with significantly higher risk of the two being noted with both low (<1 egg per week) and high (>10 eggs per week). 

That being said, coronary heart disease risk (CHD) was not significantly increased in the lower consumption group. In contrast, those in the highest group had a large increase in risk (57-122% increase relative to the group consuming 3 to less than 6 per week). The increase in CVD risk in the low group was entirely driven by stroke incidence (both hemorrhagic and ischemic, with the former having a stronger association),and conversely, the increase in CVD risk in the high group was characterized by significant increases in CHD and ischemic stroke incidence, but not hemorrhagic stroke risk. Therefore, as one might expect based on their impact on LDL/apoB, lower intake of eggs does not seem to increase CHD risk while higher intake does, although the increased stroke risk in the low intake group is still concerning. However, the author’s subgroup analysis revealed some important details that help to elucidate a possible reason for this result. 

The most notable observation was that the increased risk of CVD (as just mentioned, driven by stroke incidence) seen with lower egg intake was not present in those with an annual income over 6000 Chinese yuan (assumedly the median per capita household income), while risk increases in groups with higher intake was consistent in individuals with household incomes both over and below 6000 yuan. This would lead one to question whether the lower egg consumption is driving the risk, or if the risk increases were a result of a poor quality diet typically associated with those in, or close to, poverty, i.e., higher in refined carbohydrates, lower in fruits and vegetable, protein, and possibly calories. Consistent with this assumption, the risk within this group was also strongest in subjects with weekly intakes of fresh fruit and vegetables below 500 g, whereas the confidence interval included the null value in those with intakes over 500 g. Unfortunately there was no detailed nutrient intake data provided for the included cohorts, possibly preventing an even further demonstration that a low income-driven poor diet model is the main reason behind the ostensibly higher risk among subjects with low egg intakes. 

In the end, benefiting from a large sample, longer follow-up, broad range of egg intakes, and extensive subgroup analysis, Xia et al.’s study offered good quality data relevant to the question attempting to be answered by Godos et al.’s meta-analysis. 

CHD Incidence/Mortality

Nurses’ Health Study40

This analysis of the Nurses’ Health Study conducted by Bernstein et al. was designed to assess the impact of various protein sources on the risk of coronary heart disease. They followed a cohort of about 84000 disease-free women aged 30 to 55 for 26 years, during which 2210 myocardial infarctions and 952 deaths from CHD were recorded. The large sample, long follow up, and decent incidence rate all make this appear as a promising cohort to determine relationships between protein sources and CHD. That said, their analysis revealed that compared to the lowest intake group, those with the highest intake of eggs did not have a significantly higher risk of CHD. 

Given the collection of positive attributes this cohort possessed, it may seem reasonable to conclude this result is valid. Overall, this would be fair, although it is important to note that it shared a similar shortcoming with many of the other cohorts included in Godos et al.’s meta-analysis; the total egg intake and range of intakes were not very large at all. The highest intake group consumed a mean of 0.79 eggs daily, and the difference between high and low intakes was less than ¾ of an egg. Surprisingly, even with this major limitation, their relative risk calculation for 1 serving a day was significant in the age-adjusted model but lost significance when multivariate adjustments were performed. Though, it certainly leaned suggestively towards higher risk. In addition to this issue, nutrient intakes across quintiles of egg intake were also not provided, preventing consideration of how they may have influenced results and what eggs may have typically replaced. 

Overall, this analysis did not discover any change in CHD risk when increasing intake by a bit under ¾ of an egg daily, but the risk ratios calculated for higher (1 egg a day) intake hinted towards an increase. Despite a few impediments, the data from this cohort is still valuable given its quality otherwise. 

Greek European Prospective Investigation into Cancer and Nutrition41

The main purpose of this analysis by Dilis et al. was to determine the relationship between Mediterranean diet adherence and CHD incidence/mortality, but they also chose to observe how individual dietary components (including eggs) could influence the risk of these outcomes. Using data from just under 24000 (~12000 and ~17000) men and women aged 20-86 without pre-existing CVD or cancer who were followed up for a median of 10 years, they calculated the associated change in risk observed with a daily one standard deviation increase in the intake of various foods. Reminiscent of many of the previous cohorts, the values for the sample’s intakes make it abundantly clear that the overall amount and range of egg consumption was insufficient to draw meaningful conclusions about their impact on cardiovascular disease risk. 

The 25th percentile of men and women had intakes of 10 and 9 grams per day compared to 24 and 20 in the 75th percentile, with corresponding standard deviations of 13 and 10 grams. These standard deviations were used in their calculations, seeking to elucidate the corresponding change in CHD incidence/mortality with higher consumption despite the fact they reflect a measly difference of less than a quarter of an egg. Quite frankly, it would be absolutely incredible if such small differences in intakes reflected any notable differences in these outcomes, which makes it fairly concerning that the multivariate adjusted hazard ratio for CHD mortality with a one standard deviation increase in egg intake was the largest of all nutritional variables and hinted towards increased risk. 

However, given the nature of this analysis, the most it offers is information on the change in risk when increasing egg intake by about ¼ egg per day from a baseline already below ¼ egg, arguably of very little importance to a dose-response meta-analysis such as that carried out by Godos et al.  

Atherosclerosis Risk in Communities Study42

With this study, Haring et al. set out to find the relationship between protein type/major dietary protein sources and the risk for CHD among ~12000 cardiovascular disease and diabetes-free middle-aged adults (aged 45-64) using samples from four communities within the Atherosclerosis Risk in Communities (ARIC) cohort. After a median follow up of 22 years during which 1147 CHD events occurred, total, animal, and vegetable protein were not significantly associated with a change in CHD risk, nor were any single foods (including eggs) except for chicken. The group consuming a mean of 0.8 servings of chicken per day had a lower CHD risk than those consuming one tenth of a serving per day, which seems like quite a spurious finding, but alas this is not information relevant to the discussion at hand. As far as eggs go, intake spanned from a median of 0 to 1 per day, which is much more than can be said for other cohorts. Still, there was no association with higher risk of CHD when comparing high to low egg consumption, and results were similar for virtually all protein types and sources. 

This finding is quite puzzling given it is not consistent with those from other similar analyses of protein type and CHD, which begs the question as to what may be causing the difference. The answer is likely fairly simple and happens to be an issue that has been covered already, the final adjustment model included a moderator variable for CHD risk; total cholesterol. 

Given that increasing or decreasing total cholesterol (or more specifically LDL/apoB) is exactly how dietary components may affect the risk of CHD, adjusting for it in order to only allow for comparisons between groups with similar values would prevent the detection of risk that such changes would mediate. In addition to this, the characteristics provided for the sample stratified by protein intake were quite unsettling, with the lowest group barely even reaching the protein RDI, consuming much more carbohydrates yet less fiber, and drinking double the alcohol. 

If a large fraction of this group was included in the lowest quantile of egg consumption (the comparator) then the high intake group showing no change in CHD risk would not be very assuring. 

Due to the adjustment for serum cholesterol, it would be doubtful to see any differences in CHD risk at all, which is undoubtedly clear in this instance given the paucity of significant findings for protein type or foods. As such, the inclusion of this cohort in the meta-analysis most likely biases the effect towards null and would be detrimental to its ability to answer the original inquiry.   

Kuopio Ischaemic Heart Disease Risk Factor Study43

Virtanen et al. carried out a study on participants of a Kuopio Ischaemic Heart Disease Risk Factor cohort to determine if there was a relationship between egg or cholesterol intakes and carotid intima-media thickness (CCA-IMT) and the risk of CHD, as well as if ApoE4 phenotype modified any associations. The included cohort contained 1032 Finnish men aged 42 to 60 who were followed up for an average of about 21 years. The author’s analysis suggested that neither egg intake nor dietary cholesterol was associated with increased CHD risk or CCA-IMT. 

Alas, it was characterized by multiple issues that precluded a proper assessment of changes in risk due to the intake of the two. Aside from the fact the three tertiles encompassed a range of egg intake representing about 4/5ths of a single egg per day, the lowest median intake was 11g (compared to 52), and the lowest tertile of dietary cholesterol consumption had a median of almost 300 mg The overall dietary patterns of all three tertiles were quite frankly terrible. 

On top of the already appreciable dietary cholesterol intake in the lowest tertile, all three pertinent to both eggs and cholesterol reflected saturated fat intakes above 15% of their total energy intake and trans fat intakes over 1%. These are extraordinarily high and well beyond the thresholds of their effect on CHD risk. Further, they affect lipids and correspondingly increase the risk of CHD to a far greater degree than dietary cholesterol, and given that the relationship with their intakes and CHD are sigmoidal, once intakes are high the differences in risk seen with increases in intake of all three would be very small. However, all groups would have very high CHD rates, and no or a minor change in risk is far from helpfully informative if all groups are characterized by high risk. This is exactly what is demonstrated in the results, as CHD occurred in 20% or more of the participants in all three groups, but no difference in risk was seen when egg intake went up 4/5ths of an egg per day, which is not particularly useful information. 

Consistent with all of these comments are the almost identical lipids, blood pressures, and common carotid artery intima-media thickness (CCA-IMT) values across the entire sample population. Taking all of this into consideration, the observations in this cohort do indicate that high saturated and trans fat intake are detrimental to health, but they are far from suitable for a meta analysis looking to ascertain the impact of higher egg intake on CHD risk.        

Pan-European EPIC Cohort44

In this publication Key et al. detail their analysis of a large sample from the Pan-European EPIC cohort. They intended to delineate the potential impacts of different animal-based foods on the risk of CHD. Included in this EPIC cohort were a bit over 400000 men and women aged 35-70 from 9 European countries. After a mean follow up of 12.6 years, about 7000 CHD events or deaths had occurred. Despite boasting a large sample, a solid follow up period, and a good amount of cases, the nature of the population’s egg consumption range and the lack of detailed nutrient intake both indicate it would be unbecoming to assume the analysis accurately reflects the effect of higher egg intake on CHD risk. 

Median intakes of eggs spanned from lows of 4 g/day or 8 g/day to highs of 40 or 26, depending on whether FFQs or calibrated values from a subset that completed 24-hour recalls were used, respectively. As such, comparisons of high to low intakes would represent an increase of at most 36 g of egg (~¾ of an egg) per day, and more likely, only 18 g (about 1/3rd of an egg) per day. It should go without saying, but these are incredibly small differences and far from appropriate for determining how higher intake affects CHD risk. Furthermore, reviewing the lipid measures for each group reveals that the highest consumption group had the lowest non-HDL-cholesterol values (a good surrogate for apoB in the absence of metabolic syndrome), strongly suggesting that changes in other dietary components/nutrients likely exerted stronger beneficial impacts on lipids than the miniscule shift in egg intake. This emphasizes the importance of considering background diet.

As one might expect given these considerations, the initial analyses suggested there was a decreased risk of CHD seen with a 1 increment (20 g) increase in daily egg intake. 

Although the ability to perform a proper analysis of the effect on CHD may have been hindered appreciably, the results of this cohort do offer insight into the importance of considering reverse causality, i.e., that low egg intake may reflect active avoidance of them due to pre-existing disease or high risk. Therefore, the use of this group as a reference could bias the observed risk (or lack thereof) with increasing egg intake. 

This is exhibited by the fact that although the hazard ratio for a 20 g increase in daily egg consumption hinted that it decreased CHD risk, after removing the first four years of follow up specifically to avoid the influence of reverse-causality, this effect became non-significant and the HR shifted upwards. So in the end, this analysis did offer important information despite not being capable of serving its original purpose, determining how eggs influence CHD risk. Having completed the analysis of these two recent cohorts, a discussion of findings from protein substitution analyses follows.         

Protein Substitution Analyses

To eliminate the potential issue of failing to consider what a particular food of interest may be replacing when carrying out dose-response and/or high to low meta-analyses, many researchers have started to conduct more substitution analyses in the past decade or so. Although it was mentioned while discussing the individual cohorts in the meta-analyses above, the importance of doing this is to minimize null or even positive results that may result from comparing the food being investigated to one equally or more unhealthy. Otherwise, it could give a false impression of its overall effects on various health outcomes and conceal potential benefits that arise from replacing it with other options. Furthermore, many times they also hone in on the benefit or detriment that occurs when replacing a full serving of the food that is being pitted against another, helping to minimize the extremely prevalent issue of tiny exposure contrasts that decrease the chance of establishing an effect at all. Henceforth, by addressing some of the major problems with typical meta-analytic methods used for prospective cohorts, they can help to move towards a more complete and comprehensive characterization of the relationship between certain foods/nutrients and different outcomes.

Since protein-containing foods tend to make up a large portion of typical meals, many of these analyses have focused on the effects of substituting different protein sources for each other or other foods. This is especially valuable in the case of eggs where the data from RCTs and traditional observational research pertinent to their influence on CVD risk are discordant. The latter is especially plagued by errors that should (hopefully) be reduced with substitution analyses. Therefore, a brief review of findings from those including eggs will follow, in the hopes of bringing better clarity. 

Nurses’ Health Study and Health Professionals Follow-up Study45

Just stretching back to 2016 there have been at least 6 substitution analyses carried out using large prospective cohorts that included the replacement of eggs with other common dietary protein sources to observe potential changes in the risk of cardiovascular disease. The first of these analyses was performed by Song et al. using data from the Nurses’ Health Study and Health Professionals Follow-Up Study. These two cohorts consisted of over 130000 participants with a median age of 49 years that collected dietary data every 4 years for up to 32 years of follow-up, giving their analysis a lot of statistical power. Using this sample, they assessed the changes in risk of various outcomes seen with the replacement of animal protein sources with 3% of energy from plant protein, which they classified as bread, cereal, pasta, nuts, and legumes. Further, this percentage of energy intake would equate to around 60 calories from protein in a 2000 calorie diet, so ~2 eggs (each containing  ~30 calories from protein). Their results suggested that while replacing 3% of energy from eggs with plant protein substantially decreased the risk of all-cause, cancer, and other cause mortality, it just fell short of decreasing the risk of CVD. However, the direction of the effect unquestionably leans in that direction. 

Important to note is that this cohort (as mentioned previously) had a very low egg intake, with those in the highest animal protein consumption group only having slightly more than a median intake of 2 per week. This severely limits the confidence in their analysis that pertained to about 2 per day. Also, in the substitution model eggs were replaced by the major plant protein sources in the cohort (bread, cereal, and pasta), which aren’t necessarily healthy choices at all, especially if refined. Furthermore, adjustments also included saturated fat, which is definitely inappropriate for the analysis of CVD mortality risk considering it plays a role in increasing concentrations of a causal intermediate. If anything these characteristics should actually make the results more concerning since the saturated fat adjustment may bias results towards the null, and that refined grains are many times neutral or at worst detrimental to cardiovascular health. All that said, reductions in chronic disease risk were mostly observed in individuals with at least one unhealthy lifestyle factor. Despite these being incredibly prevalent in the United States, this may limit applicability to a healthier group. Though this analysis is of far greater quality and rigor than most of those included in the previous meta-analyses, it still had some minor pitfalls and left quite a few things to be desired. Fortunately, those to come in the following years build upon this already strong foundation. 

Japan Public Health Center–Based Prospective Cohort46

Moving forward to 2019, Budhatoki et al. carried out a very similar analysis on the changes in risk of all-cause and cause-specific mortality seen when replacing various dietary protein sources with each other. They used data from the JPHC cohort, which contained over 70000 participants from areas across Japan with a median age of just under 56 years and who were followed for a mean of 18 years. Even at baseline prior to adjustments, characteristics across quintiles of both animal and plant protein reflected a very healthy population overall and were remarkably similar (aside from protein sources, obviously), with a few minor exceptions; mostly total fat intake and refined carbohydrate intake as a percentage of calories.

Their substitution analyses of the effects of replacing different animal protein sources with 3% of energy from plant protein sources produced results very similar to Song et al.’s. Replacing 3% of energy from eggs with plant protein resulted in a significant reduction in all-cause mortality and an even stronger suggestion (though non-significant) of a reduction in CVD mortality than Song et al., only surpassed notably by replacement of red and processed meat. 

Like the analysis by Song et al., one of the main sources of plant-based protein used in their model to replace the animal based sources was cereal, followed by pulses, vegetables, and fruits. Cereals comprised over 50% of total protein, and if this were completely whole grains it wouldn’t be an issue but from the sample characteristics it is shown that 40% (of the 60% provided by carbohydrates) of their total energy intake was from refined carbohydrates. This indicates that the animal protein sources were mostly being replaced by refined carbohydrates (again, not very conducive to good cardiovascular health), yet there were still significant reductions in all-cause mortality when replacing eggs, in addition to cause-specific mortality for a few other protein sources. Also worrisome was that their highest egg intake was fairly low, with the highest consumption appearing to represent 9.5% of a total 11.7% of 2200 total calories (~260 calories) from protein, corresponding to 24 calories from egg protein, or under one egg per day. Finally, they adjusted for saturated fat, and as just discussed, this could be another important limitation. 

While improving upon a few aspects of the earlier substitution analysis, Budhatoki et al. had a few similar and some additional limitations. Regardless, they still discovered that replacing protein from eggs, even with mostly refined carbohydrates, elicited a significant reduction in all-cause mortality and almost in cardiovascular mortality despite the latter being at best neutral or potentially harmful to cardiovascular health. 

US National Institutes of Health–AARP Diet and Health Study47

In 2020, Huang et al. shared the results from their investigation into the impact of plant and animal protein on total and cause-specific mortality. Using a massive sample of over 400000 men and women with median ages of 62 from the US NIH-AARP cohort followed for 16 years, they calculated relative and absolute risk differences seen with increasing animal and plant protein intake as well as those seen with various substitutions. Not unlike Budhatoki et al., the baseline characteristics of their sample stratified by plant and animal protein intake suggested an overall healthy population with a similar distribution and minor differences in most lifestyle and dietary habits, an appreciable intake of fruits, vegetables, fiber, low rates of smoking and drinking, and a good proportion who exercised vigorously almost daily. 

Alongside observing significant reductions in the risk of almost all outcomes with higher plant protein intake (similar across almost all subgroups), their substitution analysis revealed that replacing 3% of energy intake from egg protein with plant protein resulted in substantial decreases in all cause, cancer, and CVD mortality. 

While the main sources of plant protein in this cohort were also mostly bread, cereal, and pasta (followed by “other” proteins, legumes, and nuts), the noticeably higher fiber intake suggests these included at least a fair amount of whole grains. This is likely, in part, responsible for the stark difference in risk reductions in mortality even after extensive adjustment (again including saturated fat). The one minor issue this publication shares with the other is that based on the percentages of total protein intake provided, eggs seemed to provide a small amount of overall protein, roughly representing a mean of one third to half of an egg. Still, the huge sample would have likely encompassed a wider range of intakes and allowed more accurate calculations for the changes in risk seen following replacement plant protein. 

Lifetime Risk Pooling Project48

Not long after the publication of their initial analysis on the relationship between egg intake and incident cardiovascular disease, Zhong et al. used the 6 cohorts comprising the Lifetime Risk Pooling Project to carry out a substitution analysis on dietary protein sources. To recap, these cohorts contained a total sample of just under 30,000 adults without CVD living in the United States. The median follow up was 19 years and extended up to a maximum of 31 years, providing over half a million person-years of follow up during which just under 7000 cases of CVD and 9000 deaths were recorded. Their analysis differed in the fact that not only did it calculate absolute and relative risk changes resulting from swapping single servings (instead of percent energy from protein) of the foods in question, but it also sought to determine those from replacing a serving per week, and the substitutions made were either specific protein sources or a combination. Their analysis found that replacing a single serving of eggs per week with fish, nuts, legumes, and whole grains appeared to significantly reduce the relative risk of both all-cause mortality (besides legumes) and incident CVD by 1-4%, and 30 year absolute risk up to 1%. Even more jarring, the replacement of one serving of eggs per day with fish, nuts, legumes, and whole grains elicited relative risk reductions of all-cause mortality and CVD up to 20% and 30 year absolute risk reductions ranging from 1% to around 6%, with nuts, legumes, and whole grains being the most effective, except for legumes in the case of all-cause mortality where they were overtaken by fish.  

In addition, reductions in risk associated with multiple substitutions were of an even greater magnitude, as seen below. 

In closing, as well as offering a much broader spectrum of risk calculations according to various food substitutions, another great aspect of Zhong et al.’s analysis was that an appreciable amount of the sample had at least one whole egg per day (around 8% of the 30000 participants), similar to the other protein sources that were included in their substitution models, which would lend to higher confidence in the validity of their results. Combined with the length of their follow up, the number of cases of the primary outcomes, consideration of competing risks, and careful avoidance of adjusting for causal intermediates, this made their analysis of particularly high quality. 

US National Institutes of Health–AARP Diet and Health Study Egg and Cholesterol Consumption49

Instead of focusing on multiple dietary protein sources like Huang et al. in their analysis of the US NIH-AARP Diet and Health study, Zhuang et al. used a larger sample (over half a million participants) from the same cohort to hone in on the relationship between whole egg, egg white, and dietary cholesterol consumption and all-cause/cause-specific mortality. In addition, they also carried out numerous substitution analyses specific to whole eggs. As discussed previously, this cohort contained a very large, healthy sample, encompassed a broad range of egg/dietary cholesterol consumption, had a long (16 year) mean follow up, and recorded a very large amount of all/cause-specific deaths, offering a fantastic data set for this type of investigation. 

The main finding of this analysis was that consuming just half an egg a day or 300 mg dietary cholesterol (an amount in about one and a half eggs) were associated with a significantly increased risk of all-cause mortality, CVD mortality, and cancer mortality that remained after extensive adjustment, as shown below. 

Even more intriguing were the results of their substitution analyses. These suggested that replacing one whole egg with an equivalent amount of any other protein source per day significantly reduced all-cause, CVD, and cancer mortality; with nuts, legumes, and fish eliciting the strongest effects. 

At first this might seem odd given the nature of the associations between some of the other protein sources and all-cause/CVD mortality (especially red and processed meat) in comparison to those for eggs, but upon further consideration it makes perfect sense. As indicated in the previous discourse about the cohorts included in meta-analyses observing the relationship between egg intake and CVD, almost none of them considered a high intake or exposure contrast representative of one or more eggs per day, yet those that did consistently found a considerable increase in risk. In contrast, similar investigations on the other protein sources very frequently consider higher doses and contrasts and demonstrate risk increases. This suggests the differences observed in the magnitude of the changes in risk (or lack thereof) between these foods are dependent upon these factors, not any inherent differences. However, when matching intakes of the other protein sources (red or processed meat, chicken, fish, etc.) with intake of a single egg (~70 calories and 7g of protein), both saturated fat and cholesterol intakes would almost always decrease. Although the difference may be small, these reductions would certainly compound to elicit an appreciable decrease in risk of cardiovascular disease and all-cause mortality over a longer period of time, which is in agreement with the authors findings that the associations strengthened with longer follow up. Alongside the risk of all-cause mortality being accentuated in those who were followed up longer, it was consistent across all subgroups and stronger in participants with many characteristics signifying a healthier lifestyle; non-smokers, those who were not obese, free of diabetes and hypertension, and with a higher healthy eating index score. 

Taking into account all of the strengths of this analysis, including but not limited to the very large sample, longer follow up, high incidence rate, consistency of findings across subgroups, consideration of competing risks, broad range of food intakes, extensive adjustments, and choice to consider the effect of substituting an actual serving of eggs instead of percent of protein calories it supplied, it may very well be the most thorough and highest quality publication on the topic to date, providing highly valuable data.   

Women’s Health Initiative50

Lastly, earlier in February of this year, Sun et al. published research results on the relationship between dietary protein sources and all-cause and cause-specific mortality. Like Huang et al., they also considered both the effect of individual protein sources on the risk of the outcomes of interest and the corresponding changes in risk observed when replacing them with others. For their analysis they used data from over 100000 postmenopasual women from the various components of the Women’s Health Initiative, in which just under 26000 deaths occurred during a median 18 years of follow up. Their findings with respect to the association between egg consumption and all-cause/cardiovascular mortality echoed those of Zhuang et al., higher egg intake (of a little over half an egg per day) coincided with a significantly higher risk of all-cause, cardiovascular, and cancer mortality even after extensive adjustment (again including saturated fat) and the exclusion of those who died in the first 3 years of follow up. 

Unfortunately, their substitution analyses weren’t nearly as comprehensive as some of the previous ones, but they did suggest that substitution of one egg per day with an equivalent amount of nuts would result in profound risk reductions for all three of these outcomes.

Overall, this study possessed many of the strengths that characterized the other substitution analyses, including a large sample, long follow up, large number of cases, and reasonable exposure contrast. The one minor downside was that generalizability may be limited because the population was entirely postmenopausal women. However, the results did not deviate from those observed in previous analyses containing both sexes, which offers reassurance that this might not be a concern. 

Piecing it all Together

Hopefully, the content of this article offered some clarity as to why there seems to be such a disconnect between the implications of findings regarding serum lipids from randomized controlled trials and the results of meta-analyses pertaining to the relationship between egg consumption and the risk of cardiovascular disease. 

Aside from clearing up some of the confusion surrounding this specific topic, it should also offer insight into many of the important considerations one should make when attempting to characterize a specific relationship between two variables by conducting or consulting prospective cohorts, randomized controlled trials, and most importantly, meta-analyses of the two. Any issues would only be compounded after pooling together multiple studies. While these types of studies can be incredibly useful in addressing specific research questions, one must be cautious not to take their findings at face value solely because they hold a higher position on the hierarchy of evidence. It is imperative to ensure that certain factors underpinning these investigations’ ability to even produce a reliable answer to the inquiry at hand are satisfied, otherwise the results will likely be spurious and fail to accurately portray the truth. 

Among simple considerations regarding appropriate adjustments, larger sample size and appreciable incidence of the outcome of interest, length of follow up, and the demographic being studied, many are far less conspicuous. As mentioned throughout the preceding discussion, a lot of these were common among the majority of the cohorts included in the two meta-analyses and failed to be taken into account, not unexpectedly leaving a large amount of unexplained heterogeneity for many of the results. These included the composition of participant’s background diets (especially baseline cholesterol/saturated fat intake and variation across quantiles), what the food being analyzed was replacing and that other food’s health implications, the overall outcome incidence rate as well as that of the comparator, potential for reverse causality, over-adjustment for causal intermediates, and probably the most common, low overall intakes and correspondingly small exposure contrasts. The clear pattern that emerged when evaluating the included cohorts was that in those comparing low to low egg intakes (representing differences of intake under one, or even half an egg per day), where eggs appeared to replace refined carbohydrates, and where baseline cholesterol intake and disease incidence in the comparator were high, results were mostly null and sometimes even positive. However, with such small exposure contrasts, the comparison of high intake groups to lower intake groups with high CVD rates and cholesterol intake, and replacement of foods already detrimental to cardiovascular health, it would be hasty to conclude that no effect is present or that any observed truly originated from the minimal difference in consumption. Contrastingly, a few cohorts avoided most or all of these and did suggest eggs may increase CVD risk, prompting a deeper look into other existing research on the topic. With the advent of many substitution analyses comparing various dietary protein sources and their impacts on cause-specific mortality within the past few years, much more high quality data became available to help sort out the effect of eggs on CVD.

In these large, long-term substitution analyses, researchers were much more careful to avoid many of the issues plaguing the previous prospective cohorts by including samples with a broad range of intakes, appropriately accounting for potential reverse causation, and focusing on what effect replacing eggs with other foods yielded. While the first few did have some of the aforementioned issues, the others effectively minimized the potential for such problems to occur. The initial two indicated that replacing eggs with plant-based protein sources, even despite low egg intakes and the protein sources being largely refined carbohydrates, would significantly decrease all-cause mortality. All things considered, they only hinted towards reduced cardiovascular disease mortality, which is not all that surprising given the source of the protein replacing eggs. That said, the results of the following four substitution analyses were remarkably consistent, demonstrating that increasing egg intake to over a half an egg from a baseline of  (or near) zero significantly increased all-cause and cardiovascular mortality. Furthermore, they suggested replacing a serving of eggs with fish, nuts, whole grains, legumes or egg whites per week and per day significantly reduced the risk of cardiovascular disease mortality, with the former three appearing to elicit the strongest effects. Impressively, these effects remained significant despite adjustment for saturated fat (which could bias the effect towards null) and persisted in subgroup analyses. Some even indicated that the changes in risk were of a greater magnitude among healthier participants. 

In closing, when taking all of the available evidence into consideration and focusing on results from the highest quality analyses, eggs appear to increase the risk of all-cause and cardiovascular disease mortality consistently. In many cases this effect seems to arise beyond intakes of half an egg per day (~4 per week), but it is especially uniform with intakes at or above one per day, consistent with the impacts on lipids reflected in randomized controlled feeding trials. Overall, this effect is not as large as those seen with other foods that increase CVD risk such as red/processed meat, refined carbohydrates, higher fat dairy products, and coconut/palm oil, but much of this likely owes to the disparity in intake ranges of these foods within observational studies. For those looking to absolutely minimize apoB/LDL and their corresponding CVD risk, strictly limiting or eliminating eggs would be appropriate, especially in the case that baseline intake is high. The heterogeneity observed in many meta-analyses concerning eggs and CVD risk is largely explained by a lack of rigorously designed studies. Common errors included frequent over-adjustments for causal intermediates and failure to consider reasonable exposure contrasts, replacement of low-quality foods with eggs, high baseline cholesterol intakes, and reverse causality. As such, this serves as an instance that underscores the unfortunate necessity for extreme scrutiny even when assessing ostensibly higher-quality studies that sit at or are near to the top of the evidence hierarchy.       

References: 

  1. Ference, B. A., Ginsberg, H. N., Graham, I., Ray, K. K., Packard, C. J., Bruckert, E., Hegele, R. A., Krauss, R. M., Raal, F. J., Schunkert, H., Watts, G. F., Borén, J., Fazio, S., Horton, J. D., Masana, L., Nicholls, S. J., Nordestgaard, B. G., van de Sluis, B., Taskinen, M.-R., … Catapano, A. L. (2017). Low-density lipoproteins cause atherosclerotic cardiovascular disease. 1. Evidence from genetic, epidemiologic, and clinical studies. A consensus statement from the European Atherosclerosis Society Consensus Panel. European Heart Journal, 38(32), 2459–2472. https://doi.org/10.1093/eurheartj/ehx144
  2.  Mensink, RP. Effects of saturated fatty acids on serum lipids and lipoproteins: a systematic review and regression analysis. Geneva: World Health Organization; 2016
  3. Hopkins, P. N. (1992). Effects of dietary cholesterol on serum cholesterol: A meta-analysis and review. The American Journal of Clinical Nutrition, 55(6), 1060–1070. https://doi.org/10.1093/ajcn/55.6.1060
  4. Vincent, M. J., Allen, B., Palacios, O. M., Haber, L. T., & Maki, K. C. (2019). Meta-regression analysis of the effects of dietary cholesterol intake on LDL and HDL cholesterol. The American Journal of Clinical Nutrition, 109(1), 7–16. https://doi.org/10.1093/ajcn/nqy273
  5. Sikaroudi, M. K., Soltani, S., Kolahdouz‐Mohammadi, R., Clayton, Z. S., Fernandez, M. L., Varse, F., & Shidfar, F. (2020). The responses of different dosages of egg consumption on blood lipid profile: An updated systematic review and meta-analysis of randomized clinical trials. Journal of Food Biochemistry, 44(8), e13263. https://doi.org/10.1111/jfbc.13263
  6. Brown, J. D., Buscemi, J., Milsom, V., Malcolm, R., & O’Neil, P. M. (2016). Effects on cardiovascular risk factors of weight losses limited to 5–10 %. Translational Behavioral Medicine, 6(3), 339–346. https://doi.org/10.1007/s13142-015-0353-9
  7. Godos, J., Micek, A., Brzostek, T., Toledo, E., Iacoviello, L., Astrup, A., Franco, O. H., Galvano, F., Martinez-Gonzalez, M. A., & Grosso, G. (2020). Egg consumption and cardiovascular risk: A dose–response meta-analysis of prospective cohort studies. European Journal of Nutrition. https://doi.org/10.1007/s00394-020-02345-7
  8. Krittanawong, C., Narasimhan, B., Wang, Z., Virk, H. U. H., Farrell, A. M., Zhang, H., & Tang, W. H. W. (2021). Association Between Egg Consumption and Risk of Cardiovascular Outcomes: A Systematic Review and Meta-Analysis. The American Journal of Medicine, 134(1), 76-83.e2. https://doi.org/10.1016/j.amjmed.2020.05.046
  9. Dehghan, M., Mente, A., Rangarajan, S., Mohan, V., Lear, S., Swaminathan, S., Wielgosz, A., Seron, P., Avezum, A., Lopez-Jaramillo, P., Turbide, G., Chifamba, J., AlHabib, K. F., Mohammadifard, N., Szuba, A., Khatib, R., Altuntas, Y., Liu, X., Iqbal, R., … on behalf of the PURE investigators. (2020). Association of egg intake with blood lipids, cardiovascular disease, and mortality in 177,000 people in 50 countries. The American Journal of Clinical Nutrition, 111(4), 795–803. https://doi.org/10.1093/ajcn/nqz348
  10. Kahleova, H., Crosby, L., Levin, S., & Barnard, N. D. (2018). Associations of fats and carbohydrates with cardiovascular disease and mortality — PURE and simple? The Lancet, 391(10131), 1676–1677. https://doi.org/10.1016/s0140-6736(18)30805-5
  11. Lorkowski, S., Richter, M., Linseisen, J., & Watzl, B. (2018). Associations of fats and carbohydrates with cardiovascular disease and mortality — PURE and simple? The Lancet, 391(10131), 1678–1679. https://doi.org/10.1016/s0140-6736(18)30800-6
  12. Li, D. (2017). Is it really good for you to eat fat as much as you could? Science China Life Sciences, 61(3), 363–364. doi:10.1007/s11427–017–9194–3
  13. Mann, J., Meerpohl, J., Nishida, C., McLean, R., & Te Morenga, L. (2018). Associations of fats and carbohydrates with cardiovascular disease and mortality — PURE and simple? The Lancet, 391(10131), 1676. https://doi.org/10.1016/s0140-6736(18)30804-3
  14. Cao, X.-P., Tan, C.-C., & Yu, J.-T. (2018). Associations of fats and carbohydrates with cardiovascular disease and mortality — PURE and simple? The Lancet, 391(10131), 1679–1680. https://doi.org/10.1016/s0140-6736(18)30793-1
  15. Djoussé, L., Ho, Y.-L., Nguyen, X.-M. T., Quaden, R. M., Gagnon, D. R., Gaziano, J. M., & Cho, K. (2020). Egg consumption and risk of coronary artery disease in the Million Veteran Program. Clinical Nutrition (Edinburgh, Scotland), 39(9), 2842–2847. https://doi.org/10.1016/j.clnu.2019.12.017
  16. Qin, C., Lv, J., Guo, Y., Bian, Z., Si, J., Yang, L., Chen, Y., Zhou, Y., Zhang, H., Liu, J., Chen, J., Chen, Z., Yu, C., & Li, L. (2018). Associations of egg consumption with cardiovascular disease in a cohort study of 0.5 million Chinese adults. Heart, 104(21), 1756–1763. https://doi.org/10.1136/heartjnl-2017-312651
  17. Xu, L., Lam, T. H., Jiang, C. Q., Zhang, W. S., Zhu, F., Jin, Y. L., Woo, J., Cheng, K. K., & Thomas, G. N. (2019). Egg consumption and the risk of cardiovascular disease and all-cause mortality: Guangzhou Biobank Cohort Study and meta-analyses. European Journal of Nutrition, 58(2), 785–796. https://doi.org/10.1007/s00394-018-1692-3
  18. Guo, J., Hobbs, D. A., Cockcroft, J. R., Elwood, P. C., Pickering, J. E., Lovegrove, J. A., & Givens, D. I. (2018). Association between egg consumption and cardiovascular disease events, diabetes and all-cause mortality. European Journal of Nutrition, 57(8), 2943–2952. https://doi.org/10.1007/s00394-017-1566-0
  19.  Djoussé, L., & Gaziano, J. M. (2008). Egg Consumption and Cardiovascular Disease and Mortality The Physicians’ Health Study. The American Journal of Clinical Nutrition, 87(4), 964–969.
  20. Qureshi, A. I., Suri, M. F. K., Ahmed, S., Nasar, A., Divani, A. A., & Kirmani, J. F. (2006). Regular egg consumption does not increase the risk of stroke and cardiovascular diseases. Medical Science Monitor, 13(1), CR1–CR8.
  21. Hu, F. B., Stampfer, M. J., Rimm, E. B., Manson, J. E., Ascherio, A., Colditz, G. A., Rosner, B. A., Spiegelman, D., Speizer, F. E., Sacks, F. M., Hennekens, C. H., & Willett, W. C. (1999). A prospective study of egg consumption and risk of cardiovascular disease in men and women. JAMA, 281(15), 1387–1394. https://doi.org/10.1001/jama.281.15.1387
  22. Li Yanping, Hruby Adela, Bernstein Adam M., Ley Sylvia H., Wang Dong D., Chiuve Stephanie E., Sampson Laura, Rexrode Kathryn M., Rimm Eric B., Willett Walter C., & Hu Frank B. (2015). Saturated Fats Compared With Unsaturated Fats and Sources of Carbohydrates in Relation to Risk of Coronary Heart Disease. Journal of the American College of Cardiology, 66(14), 1538–1548. https://doi.org/10.1016/j.jacc.2015.07.055
  23. Jakobsen, M. U., Dethlefsen, C., Joensen, A. M., Stegger, J., Tjønneland, A., Schmidt, E. B., & Overvad, K. (2010). Intake of carbohydrates compared with intake of saturated fatty acids and risk of myocardial infarction: Importance of the glycemic index. The American Journal of Clinical Nutrition, 91(6), 1764–1768. https://doi.org/10.3945/ajcn.2009.29099
  24. Zhong, V. W., Van Horn, L., Cornelis, M. C., Wilkins, J. T., Ning, H., Carnethon, M. R., Greenland, P., Mentz, R. J., Tucker, K. L., Zhao, L., Norwood, A. F., Lloyd-Jones, D. M., & Allen, N. B. (2019). Associations of Dietary Cholesterol or Egg Consumption With Incident Cardiovascular Disease and Mortality. JAMA, 321(11), 1081. https://doi.org/10.1001/jama.2019.1572
  25. Jang, J., Shin, M.-J., Kim, O. Y., & Park, K. (2018). Longitudinal association between egg consumption and the risk of cardiovascular disease: Interaction with type 2 diabetes mellitus. Nutrition & Diabetes, 8(1), 1–9. https://doi.org/10.1038/s41387-018-0033-1
  26. Díez-Espino, J., Basterra-Gortari, F. J., Salas-Salvadó, J., Buil-Cosiales, P., Corella, D., Schröder, H., Estruch, R., Ros, E., Gómez-Gracia, E., Arós, F., Fiol, M., Lapetra, J., Serra-Majem, L., Pintó, X., Babio, N., Quiles, L., Fito, M., Marti, A., & Toledo, E. (2017). Egg consumption and cardiovascular disease according to diabetic status: The PREDIMED study. Clinical Nutrition, 36(4), 1015–1021. https://doi.org/10.1016/j.clnu.2016.06.009
  27. Scrafford, C. G., Tran, N. L., Barraj, L. M., & Mink, P. J. (2011). Egg consumption and CHD and stroke mortality: A prospective study of US adults. Public Health Nutrition, 14(2), 261–270. https://doi.org/10.1017/S1368980010001874
  28. Zazpe, I., Beunza, J., Bes-Rastrollo, M. et al. Egg consumption and risk of cardiovascular disease in the SUN Project. Eur J Clin Nutr 65, 676–682 (2011). https://doi.org/10.1038/ejcn.2011.30
  29. Houston, D. K., Ding, J., Lee, J. S., Garcia, M., Kanaya, A. M., Tylavsky, F. A., Newman, A. B., Visser, M., & Kritchevsky, S. B. (2011). Dietary fat and cholesterol and risk of cardiovascular disease in older adults: The Health ABC Study. Nutrition, Metabolism and Cardiovascular Diseases, 21(6), 430–437. https://doi.org/10.1016/j.numecd.2009.11.007
  30. Nakamura, Y., Iso, H., Kita, Y., Ueshima, H., Okada, K., Konishi, M., Inoue, M., Tsugane, S., & Group, J. P. H. C. prospective study. (2006). Egg consumption, serum total cholesterol concentrations and coronary heart disease incidence: Japan Public Health Center-based prospective study. British Journal of Nutrition, 96(5), 921–928. https://doi.org/10.1017/BJN20061937
  31. Burke, V., Zhao, Y., Lee, A. H., Hunter, E., Spargo, R. M., Gracey, M., Smith, R. M., Beilin, L. J., & Puddey, I. B. (2007). Health-related behaviours as predictors of mortality and morbidity in Australian Aborigines. Preventive Medicine, 44(2), 135–142. https://doi.org/10.1016/j.ypmed.2006.09.008
  32. Trichopoulou, A., Psaltopoulou, T., Orfanos, P., & Trichopoulos, D. (2006). Diet and physical activity in relation to overall mortality amongst adult diabetics in a general population cohort. Journal of Internal Medicine, 259(6), 583–591. https://doi.org/10.1111/j.1365-2796.2006.01638.x
  33. Goldberg, S., Gardener, H., Tiozzo, E., Kuen, C. Y., Elkind, M. S. V., Sacco, R. L., & Rundek, T. (2014). Egg consumption and carotid atherosclerosis in the Northern Manhattan Study. Atherosclerosis, 235(2), 273–280. https://doi.org/10.1016/j.atherosclerosis.2014.04.019
  34. Larsson, S. C., Åkesson, A., & Wolk, A. (2015). Egg consumption and risk of heart failure, myocardial infarction, and stroke: Results from 2 prospective cohorts. The American Journal of Clinical Nutrition, 102(5), 1007–1013. https://doi.org/10.3945/ajcn.115.119263
  35. Farvid, M. S., Malekshah, A. F., Pourshams, A., Poustchi, H., Sepanlou, S. G., Sharafkhah, M., Khoshnia, M., Farvid, M., Abnet, C. C., Kamangar, F., Dawsey, S. M., Brennan, P., Pharoah, P. D., Boffetta, P., Willett, W. C., & Malekzadeh, R. (2017). Dietary Protein Sources and All-Cause and Cause-Specific Mortality: The Golestan Cohort Study in Iran. American Journal of Preventive Medicine, 52(2), 237–248. https://doi.org/10.1016/j.amepre.2016.10.041
  36. Zamora-Ros, R., Cayssials, V., Cleries, R., Redondo, M. L., Sánchez, M.-J., Rodríguez-Barranco, M., Sánchez-Cruz, J.-J., Mokoroa, O., Gil, L., Amiano, P., Navarro, C., Chirlaque, M. D., Huerta, J. M., Barricarte, A., Ardanaz, E., Moreno-Iribas, C., & Agudo, A. (2019). Moderate egg consumption and all-cause and specific-cause mortality in the Spanish European Prospective into Cancer and Nutrition (EPIC-Spain) study. European Journal of Nutrition, 58(5), 2003–2010. https://doi.org/10.1007/s00394-018-1754-6
  37. van den Brandt, P. A., Goldbohm, R. A., van ’t Veer, P., Volovics, A., Hermus, R. J., & Sturmans, F. (1990). A large-scale prospective cohort study on diet and cancer in The Netherlands. Journal of Clinical Epidemiology, 43(3), 285–295. https://doi.org/10.1016/0895-4356(90)90009-e
  38. Drouin-Chartier, J.-P., Chen, S., Li, Y., Schwab, A. L., Stampfer, M. J., Sacks, F. M., Rosner, B., Willett, W. C., Hu, F. B., & Bhupathiraju, S. N. (2020). Egg consumption and risk of cardiovascular disease: Three large prospective US cohort studies, systematic review, and updated meta-analysis. BMJ, 368, m513. https://doi.org/10.1136/bmj.m513
  39. Xia, X., Liu, F., Yang, X., Li, J., Chen, J., Liu, X., Cao, J., Shen, C., Yu, L., Zhao, Y., Wu, X., Zhao, L., Li, Y., Huang, J., Lu, X., & Gu, D. (2020). Associations of egg consumption with incident cardiovascular disease and all-cause mortality. Science China Life Sciences, 63(9), 1317–1327. https://doi.org/10.1007/s11427-020-1656-8
  40. Bernstein, A. M., Sun, Q., Hu, F. B., Stampfer, M. J., Manson, J. E., & Willett, W. C. (2010). Major dietary protein sources and risk of coronary heart disease in women. Circulation, 122(9), 876–883. https://doi.org/10.1161/CIRCULATIONAHA.109.915165
  41. Dilis, V., Katsoulis, M., Lagiou, P., Trichopoulos, D., Naska, A., & Trichopoulou, A. (2012). Mediterranean diet and CHD: The Greek European Prospective Investigation into Cancer and Nutrition cohort. British Journal of Nutrition, 108(4), 699–709. https://doi.org/10.1017/S0007114512001821
  42. Haring, B., Misialek, J. R., Rebholz, C. M., Petruski-Ivleva, N., Gottesman, R. F., Mosley, T. H., & Alonso, A. (2015). Association of dietary protein consumption with incident silent cerebral infarcts and stroke: The ARIC Study. Stroke; a Journal of Cerebral Circulation, 46(12), 3443–3450. https://doi.org/10.1161/STROKEAHA.115.010693
  43. Virtanen, J. K., Mursu, J., Virtanen, H. E., Fogelholm, M., Salonen, J. T., Koskinen, T. T., Voutilainen, S., & Tuomainen, T.-P. (2016). Associations of egg and cholesterol intakes with carotid intima-media thickness and risk of incident coronary artery disease according to apolipoprotein E phenotype in men: The Kuopio Ischaemic Heart Disease Risk Factor Study. The American Journal of Clinical Nutrition, 103(3), 895–901. https://doi.org/10.3945/ajcn.115.122317
  44. Key Timothy J., Appleby Paul N., Bradbury Kathryn E., Sweeting Michael, Wood Angela, Johansson Ingegerd, Kühn Tilman, Steur Marinka, Weiderpass Elisabete, Wennberg Maria, Lund Würtz Anne Mette, Agudo Antonio, Andersson Jonas, Arriola Larraitz, Boeing Heiner, Boer Jolanda M.A., Bonnet Fabrice, Boutron-Ruault Marie-Christine, Cross Amanda J., … Danesh John. (2019). Consumption of Meat, Fish, Dairy Products, and Eggs and Risk of Ischemic Heart Disease. Circulation, 139(25), 2835–2845. https://doi.org/10.1161/CIRCULATIONAHA.118.038813
  45. Song, M., Fung, T. T., Hu, F. B., Willett, W. C., Longo, V. D., Chan, A. T., & Giovannucci, E. L. (2016). Association of Animal and Plant Protein Intake With All-Cause and Cause-Specific Mortality. JAMA Internal Medicine, 176(10), 1453. https://doi.org/10.1001/jamainternmed.2016.4182
  46. Budhathoki, S., Sawada, N., Iwasaki, M., Yamaji, T., Goto, A., Kotemori, A., Ishihara, J., Takachi, R., Charvat, H., Mizoue, T., Iso, H., Tsugane, S., & for the Japan Public Health Center–based Prospective Study Group. (2019). Association of Animal and Plant Protein Intake With All-Cause and Cause-Specific Mortality in a Japanese Cohort. JAMA Internal Medicine, 179(11), 1509. https://doi.org/10.1001/jamainternmed.2019.2806
  47. Huang, J., Liao, L. M., Weinstein, S. J., Sinha, R., Graubard, B. I., & Albanes, D. (2020). Association Between Plant and Animal Protein Intake and Overall and Cause-Specific Mortality. JAMA Internal Medicine, 180(9), 1173–1184. https://doi.org/10.1001/jamainternmed.2020.2790
  48. Zhong, V. W., Allen, N. B., Greenland, P., Carnethon, M. R., Ning, H., Wilkins, J. T., Lloyd-Jones, D. M., & Van Horn, L. (2021). Protein foods from animal sources, incident cardiovascular disease and all-cause mortality: A substitution analysis. International Journal of Epidemiology, 50(1), 223–233. https://doi.org/10.1093/ije/dyaa205
  49. Zhuang, P., Wu, F., Mao, L., Zhu, F., Zhang, Y., Chen, X., Jiao, J., & Zhang, Y. (2021). Egg and cholesterol consumption and mortality from cardiovascular and different causes in the United States: A population-based cohort study. PLOS Medicine, 18(2), e1003508. https://doi.org/10.1371/journal.pmed.1003508
  50. Sun Yangbo, Liu Buyun, Snetselaar Linda G., Wallace Robert B., Shadyab Aladdin H., Kroenke Candyce H., Haring Bernhard, Howard Barbara V., Shikany James M., Valdiviezo Carolina, & Bao Wei. (2021). Association of Major Dietary Protein Sources With All‐Cause and Cause‐Specific Mortality: Prospective Cohort Study. Journal of the American Heart Association, 10(5), e015553. https://doi.org/10.1161/JAHA.119.015553

SUBSCRIBE TO EMAIL LIST

Related Deep Dives

Bodyweight Control
Screenshot 2022-05-18 at 19.13.40

Eating Behaviour and Obesity are Choices? When Science Meets Philosophy

Disease Prevention
Screenshot 2021-11-27 at 15.59.15

Merging Reductionism and Holism in Nutrition

Disease Prevention
Screenshot 2021-10-11 at 16.38.54

Fibre and Constipation: Correcting Paul Mason Misinformation