Introduction

The concept that diet, serum cholesterol and cardiovascular disease are causally related gave rise to the diet-heart hypothesis nearly 70 years ago. This hypothesis postulates that reducing dietary saturated fat reduces serum cholesterol, thereby reducing the risk of cardiovascular disease. Today, this concept has been transformed from a hypothesis into public health policy as current guidelines recommend reducing the intake of dietary saturated fat.1 Not all practitioners agree, however, and a reappraisal of the evidence may help resolve this controversy.

An abbreviated history

Dr Ancel Keys first proposed the diet-heart hypothesis in the 1950s. Several years later, he published the Seven Countries Study that reported a strong correlation between dietary fat and coronary mortality in seven countries.2 His hypothesis rapidly gained support and by 1977 the US Senate Select Committee on Nutrition and Human Needs formally recommended that Americans should reduce their consumption of total and saturated fat.3 Many scientists disagreed and a vitriolic debate soon followed.4 Presently, the American Heart Association recommends ‘lowering intake of saturated fat and replacing it with unsaturated fats, especially polyunsaturated fats’ to reduce cardiovascular disease.5

Fat and cholesterol

Low-density lipoprotein-cholesterol is a cholesterol-containing multi-molecular complex often targeted for reduction because of its central role in atherosclerosis, yet cholesterol is essential for life as a key constituent of cell membranes and bile acids and as a precursor of vitamin D and steroid hormones. Serum cholesterol levels reflect both dietary fat intake and the synthesis of cholesterol within the body. Most animal fats are saturated fats and tend to raise serum cholesterol levels. Unsaturated fats generally do not raise cholesterol levels and include monounsaturated (omega-7 and omega-9) and polyunsaturated (omega-3 and omega-6) fats. Industrially produced Trans fats raise serum cholesterol and increase the risk of cardiovascular disease. Clinical trials that reduce the intake of saturated fat replace this macronutrient with unsaturated fat, carbohydrate or protein to maintain the caloric balance.

Randomised controlled trials

Dietary randomised controlled trials (RCTs) are fraught with methodological problems and difficult to conduct. Despite these limitations, we believe they remain the best means for validating or rejecting a medical hypothesis. We identified 28 RCTs of diet lasting at least 1 year and reporting cardiovascular and/or mortality outcomes (table 1). Some trials had significant problems that might undermine their conclusions. For example, the Finnish Mental Hospital Study was excluded from a 2015 Cochrane meta-analysis (table 2) because it randomised subjects based on their hospital location (cluster randomisation).6 7 Other studies employed multiple simultaneous interventions making it nearly impossible to determine the impact of dietary intervention alone. Diet and reinfarction trial (DART)-2 had major methodological problems including interruption for 1 year.8 Some experts have discredited PREDIMED  (Prevención con Dieta Mediterránea) after it was retracted and later republished although with nearly identical results.9 Of the remaining 22 RCTs, the overarching observation is the paucity of benefit. Only two trials reported a mortality benefit (Lyon and DART-fish) and both of these trials reported no change in serum cholesterol levels.10 11 Notably, DART-fish simply supplemented the control diet with fish and the Lyon trial utilised a Mediterranean diet that replaced saturated fat with a combination of omega-3 and omega-9 fatty acids associated with a reduction of omega-6 fatty acids.12 In total, 11 of the 22 trials reported statistically significant reductions in serum cholesterol, but none of these reported a mortality benefit and only two reported a reduction in cardiovascular events. Of the eight RCTs that specifically replaced saturated fat with various combinations of omega-3 and omega-6 polyunsaturated fats, none reported a mortality benefit and only two reported a reduction in cardiovascular events. Two RCTs reported increased mortality and/or cardiovascular events with cholesterol reduction.

No RCT is perfect and we should acknowledge the marked variability of patient characteristics, population sizes, trial durations, interventions and the degree of cholesterol reduction reported in these trials. Furthermore, the replacement macronutrient may be of particular importance in assessing these studies.13 Nevertheless, it appears that the greatest benefit was seen in dietary RCTs that did not lower serum cholesterol and the vast majority of cholesterol-lowering dietary RCTs, including those replacing saturated fat with polyunsaturated fat, were ineffective in reducing cardiovascular events or mortality.

Meta-analyses

We identified 17 meta-analyses that broadly examined the relationship of dietary fat to cardiovascular disease and/or mortality.

Of the 11 meta-analyses with mortality outcomes, only two reported a mortality benefit. Of the 14 with cardiovascular event outcomes six reported benefit and eight did not. Eight meta-analyses included only RCTs that replaced saturated fat with polyunsaturated fat. None of these reported a mortality benefit and only two reported a reduction in cardiovascular events. Only the Schwingshackl meta-analysis examined the replacement of saturated fat with polyunsaturated fat in secondary prevention and found no reduction in mortality or cardiovascular events.14 To summarise, the results of most meta-analyses do not support the diet-heart hypothesis or the recommendation to replace saturated fat with polyunsaturated fat.

Unintended consequences

Low-fat foods are commonly marketed as being heart healthy, but to make low-fat foods more palatable the food industry has developed low-fat foods that are often high in refined carbohydrates. Paradoxically, the increased consumption of refined carbohydrates in the USA has been linked to the epidemic of diabetes that can potentially lead to heart disease.15

Conclusions

The preponderance of evidence indicates that low-fat diets that reduce serum cholesterol do not reduce cardiovascular events or mortality. Specifically, diets that replace saturated fat with polyunsaturated fat do not convincingly reduce cardiovascular events or mortality. These conclusions stand in contrast to current opinion.5 There are several possible explanations. Foremost, we must consider that the diet-heart hypothesis is invalid or requires modification.16 Moreover, some experts may selectively cite evidence that validates their own viewpoint while disregarding evidence to the contrary, a behaviour called confirmation bias.17 Others have noted the limitations of targeting a single risk factor in a multifactorial disease.18 Alternatively, the degree and/or duration of cholesterol reduction achieved with diet may simply be inadequate. Finally, the concept that one can simply replace one macronutrient with another to prevent cardiovascular disease risks oversimplifying an extremely complex disease process.