Carbohydrates, not animal fats, linked to heart disease across 42 European countries

A study of 42 European countries found lower cardiovascular disease and mortality among countries that consumed more fats and animal protein. Higher cardiovascular mortality was linked to carbohydrate consumption. Another nail in the coffin for the diet-heart hypothesis?

Read time Key points: 4 minutes (700 words plus several tasty graphs); Additional analysis: +4 minutes (+700 words)

Controversy over the diet-heart hypothesis rages on: the role of saturated fat, animal products, LdL vs HdL cholesterol levels, statin drugs, low-carb vs low-fat diets… So here is another piece of the jigsaw puzzle that reveals some pretty striking correlations.

Study design

The study by a researchers in the Czech Republic was published in Food and Nutrition Research last year but gained little media attention so few in the nutrition community seem to have heard of it. It examined the mean consumption of 62 food items from the FAOSTAT database (1993–2008) and compared these with the actual statistics of five cardio-vascular disease (CVD) indicators across 42 European countries.

The FAOSTAT database compiles the total food consumption by category at a national level. Although it tells us noting about any individuals consumption this data is a robust and reliable method of assessing each countries consumption over all. Its use in this study provides a pretty objective picture and avoids the pitfalls and bias involved in most epidemiological studies which rely on participants accurately recalling personal food consumption over a given period – a notoriously inaccurate affair. So what did they find?

Key findings

  • Cholesterol levels were tightly correlated to the consumption of animal fats and proteins – Countries consuming more fat and protein from animal sources had higher incidence of raised cholesterol
  • Raised cholesterol correlated negatively with CVD risk – Countries with higher levels of raised cholesterol had fewer cases of CVD deaths and a lower incidence of CVD risk factors
  • Carbohydrates correlated positively with CVD risk – the more carbohydrates consumed (and especially those with high GI such as starches) the more CVD
  • Fat and Protein correlated negatively with CVD risk – Countries consuming more fat and protein from animal and plant sources had less CVD. The authors speculate that this is because increasing fat and protein in the diet generally displaces carbohydrates.

The study includes a huge amount of data analysis, which for a sub set of countries spanned three decades, allowing tracking of these correlations over time. Interestingly, in the 1980s there was a brief period when CVD rates correlated positively with fat intake, which the authors speculate may have led to the reinforcing of the saturated fat and heart disease hypothesis, but these trends have long since reversed.

Cholesterol levels

The strongest predictor of raised cholesterol that came from this analysis was consumption of animal fat and animal protein – which includes cheese and meat. (Only graph for men shown, but graph for women is similar)


Note, how closely the points lie along the line. A correlation (r value) of 0.92 indicates that 85% of the variation in raised cholesterol (in men) can be predicted by their consumption of animal fat and protein (primarily meat and cheese)

On the other hand, carbohydrates from cereals and potatoes (i.e. starches) were inversely correlated with raised cholesterol.


The authors state that these results are expected, based on previous studies:

This is in accordance with the meta-analyses of clinical trials, which show that saturated animal fat is the major trigger of raised cholesterol

So far, so good. But how do these correlations translate into cardiovascular health?

CVD Mortality and Raised Cholesterol

As the graph below shows, and contrary to received wisdom, the higher the cholesterol, the lower the CVD mortality:


Whilst many people would find this surprising, being the exact opposite of what we are all told, the authors explain:

The negative relationship between raised cholesterol and CVD may seem counterintuitive, but it is not at variance with the available evidence. The largest of the recent worldwide meta-analyses dealing with cholesterol and CVD risk observed a positive relationship between raised cholesterol and CVD mortality at younger ages, but this association gradually started to reverse in seniors, where the number of deaths is the highest.

CVD Mortality and Total fat and Animal protein

The strongest negative correlation was between Total fat and Animal protein and CVD mortality, especially in women. i.e. countries with lower rates of CVD mortality (in women) consumed more fat and animal protein: 

This is extraordinary, and again the reverse of what we are usually told. At the top left: France; The so-called ‘French paradox’ – high consumption of saturated fat and animal products, but low levels of CVD – really does not look that paradoxical in this study; in fact it’s pretty much where you would expect based on all of the other countries in Europe.

CVD Mortality and Carbohydrates

As expected from all the above, countries with higher rates of CVD mortality consumed more carbohydrates (including alcohol):

CVD Risk Factors

The prevalence of certain CVD risk factors (raised blood pressure & raised blood glucose) was assessed for each country. Carbohydrates (primarily from cereals) provided the strongest correlation with these two risk factors:


Strong positive correlations were found between total energy from carbohydrates (‘CA energy’ in the graphs above) and prevalence of raised blood pressure (left) and raised blood glucose (right)

That’s the headline findings. If you like the details continue below!



The most obvious potential confounder in a study like this is economic. In each of the graphs above, we see the higher levels of CVD mortality among the poorer Eastern European countries, and lower levels among the Northern and Western European countries.

The argument would go something like this: countries with higher GDPs can afford more meat and fat, but also better health care. Hence, the reduction in CVD seen in the graphs above may be due to health spending not diet.

Indeed the study found that Health expenditure in each country was tightly correlated to CVD mortality:

So does this mean that the beneficial associations of CVD mortality with fat and animal protein on the one hand, and the negative association of carbohydrates on the other are simply markers for health expenditure? It seems unlikely.

First, it is notable that the health expenditure graph forms a curve, yet the dietary graphs are linear, suggesting that they have at least some independent association and are not simply surrogate markers of health expenditure.

Furthermore, if the health spending graph is considered in two sections, (Western v Eastern European countries) we see that there is little correlation. Among the 21 countries spending below US$ 2000 per capita there is only the vaguest of correlations between CVD mortality and health expenditure. Whereas for the remaining 21 countries spending above US$ 2000 per capita shows almost no effect on CVD mortality.

Finally, CVD risk factors – raised blood pressure and raised blood glucose – which are indicative of CVD prevalence are both correlated tightly to carbohydrate consumption (see below).

Health expenditure – the main confounder in this study – is clearly related to CVD mortality, but its influence is not apparent in the case of raised blood pressure or blood glucose, which depend on the individual lifestyle. It is also difficult to imagine that health expenditure would be able to completely reverse the connection between nutrition and all the selected CVD indicators.

Despite attempting to correct for confounders using multiple statistical models the authors concluded that the dietary trends remained and although such correlations cannot prove causation, they do disprove the opposite: Saturated fat consumption does not increase heart disease.

Irrespective of the possible limitations of the ecological study design, the undisputable finding of our paper is the fact that the highest CVD prevalence can be found in countries with the highest carbohydrate consumption, whereas the lowest CVD prevalence is typical of countries with the highest intake of fat and protein.


…the strong ecological relationship between CVD prevalence and carbohydrate consumption is a serious challenge to the current concepts of the aetiology of CVD.


The authors suggest this is not as unexpected as it might seem:

These findings strikingly contradict the traditional ‘saturated fat hypothesis’, but in reality, they are compatible with the evidence accumulated from observational studies that points to both high glycaemic index and high glycaemic load (the amount of consumed carbohydrates × their glycaemic index) as important triggers of CVDs. The highest glycaemic indices (GI) out of all basic food sources can be found in potatoes and cereal products, which also have one of the highest food insulin indices (FII) that betray their ability to increase insulin levels.

Other Associations

In this post I have focused on the macronutrients: carbohydrates, fats and animal proteins, as these were the main story of this study, but there were other interesting points that were revealed which I will summarise briefly here:

Foods associated with lower CVD Foods associated with higher CVD
Oranges Distilled beverages
Wine Sunflower oil*
High fat dairy, esp. cheese Onions*
Tree nuts / olives
Low glycaemic vegetables
Fish and Seafood

*it is possible that both these food items mirror an environment that is deficient in some important factors correlating negatively with CVD risk

End note

The authors point out that a previous study looking at cancer risk across European countries found the opposite trends. With higher cancer incidence in North and Western Europe, Lower in Eastern.

The best example of this health paradox is… that of French men, who have the lowest rates of CVD mortality in Europe, but the highest rates of cancer incidence. In other words, cancer and CVDs appear to express two extremes of a fundamental metabolic disbalance that is related to factors such as cholesterol and IGF-1 (insulin-like growth factor).

But that’s a story for another day.


Food consumption and the actual statistics of cardiovascular diseases: an epidemiological comparison of 42 European countries. Grasgruber et al, Food and nutrition research, Sep 2016 [PubMed] [PMC full text]




  1. “Ancel Benjamin Keys,
    Who studied nutritional disease
    Said if results are contrarian,
    I’ll just bury ’em
    -did Ancel Benjamin Keys”

    In My Humble Opinion, sums it up.

    (- with apologies to the Original Author if I’ve mangled his limerick)

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.