Friday, 28 March 2014

Clearing Up The Confusion Surrounding Saturated Fat

In 2010, Siri-Tarino and colleagues published a meta-analysis of prospective cohort studies evaluating the association between dietary saturated fat and cardiovascular disease in the American Journal of Clinical Nutrition.1 Based on the results of this meta-analysis, these researchers concluded that there was insufficient evidence from prospective cohort studies to conclude that dietary saturated fat increases the risk of coronary heart disease. However, a number of prominent diet-heart researchers identified many serious flaws and omissions in this meta-analysis that cast doubt on the validity of these researchers conclusions.2 3 4 5 6

More recently, Chowdhury and colleagues published a separate meta-analysis in the Annals of Internal Medicine, and reached similar conclusions to that of Siri-Tarino and colleagues regarding the association between saturated fat and coronary heart disease.7 Unfortunately, this meta-analysis also failed to sufficiently address a number of important limitations that it shares with the meta-analysis by Siri-Tarino and colleagues. Furthermore, in this meta-analysis, although positively, but not significantly associated in the random-effects model, both dietary and total circulating concentrations of saturated fat were associated with a small, but statistically significant increased risk of coronary heart disease in the fixed effects model (RR=1.04 [95% CI, 1.01, 1.07] and RR=1.13 [95% CI, 1.03-1.25], respectively). These significant findings were however ignored in the conclusions of this study. Nevertheless, the media and proponents of popular Low-Carb and Paleo diets have repeatedly cited these meta-analyses as evidence to support a diet rich in saturated fat. 


Saturated Fat and Coronary Heart Disease Mortality


In the editorial to the Siri-Tarino meta-analysis, Jeremiah Stamler noted that saturated fat intake was more strongly associated with fatal than non-fatal incidence of coronary heart disease. Stamler calculated that based on the 11 studies included in the meta-analysis which provided estimates specifically for fatal cases, saturated fat was associated with a 32% increased risk of death from coronary heart disease, when weighted by person-years of exposure.2 Siri-Tarino and colleagues noted this concern in a follow-up paper, but instead downplayed these findings by asserting that in their own analysis of only 7 studies, saturated fat intake was associated with only a borderline significant 18% increased risk of death from coronary heart disease, when using the random effects model (RR=1.18 [95% CI 0.99-1.42]).8 Similarly, in the more recent meta-analysis, Chowdhury and colleagues found that in their sub-analysis of only 9 studies, saturated fat intake was associated with a borderline significant 7% increased risk of death from coronary heart disease (RR=1.07 [95% CI, 1.00-1.13]).7

It can be deduced by the estimates and the sample size of these sub-analyses by both Siri-Tarino and Chowdhury, that only the studies that provided estimates specifically only for fatal cases were included. Therefore these sub-analyses excluded studies that provided estimates for fatal cases in additional to that of total incidence of coronary heart disease. In total, 14 prospective cohort studies provided estimates for death from coronary heart disease, of which 3 were not included in the original meta-analysis by Siri-Tarino and colleagues.1 9 10 11 12 13 14 15 16 17 18 19 20 21 22

The exclusion of several studies in these analyses warrants a reanalysis of the studies evaluating the association between saturated fat and the risk of death from coronary heart disease. I therefore performed a meta-analysis including all 14 studies for which estimates were available specifically for death from coronary heart disease. Similar to the methods of Siri-Tarino and colleagues, I chose to compare extreme quantiles of saturated fat intake where possible. However, for the studies which estimates were provided as either a 1% increase of energy, or as a 1-unit increase, the estimates were transformed to represent roughly a 5% increase in energy from saturated fat, as this was similar to the difference for high vs low quantiles of intake in the other studies included in this meta-analysis. In order to ensure that the methods used for the statistical analysis were consistent with that used by Siri-Tarino and colleagues, I performed the meta-analysis in Review Manager (from The Cochrane Collaboration), and pooled the estimates using the random effects model for both within-study and between-study variation. Similarly, risk ratios and 95% confidence intervals were log transformed to derive the corresponding standard error for beta-coefficients by using Greenland’s formula.23 Otherwise, the exact P-value was used where available to derive the corresponding standard error.

In a meta-analysis including 14 studies, dietary saturated fat intake was associated with a statistically highly significant 24% increased risk of death from coronary heart disease (Fig. 1). Similarly, for the 11 studies included in the Siri-Tarino meta-analysis, saturated fat was associated with a statistically highly significant 26% increased risk of death from coronary heart disease (RR=1.26 [95% CI, 1.14-1.40]). 

FIGURE 1. Risk ratios and 95% CIs for fully adjusted random-effects models examining associations between saturated fat intake in relation to coronary heart disease mortality. ¹Studies that included adjustments for serum or LDL cholesterol. SAT, saturated fat intake.

It is well established that saturated fat raises serum and LDL cholesterol, and that these blood lipids increase the risk of coronary heart disease.24 25 26 27 However, in this meta-analysis, almost 40% of the weight was derived from studies that controlled for either serum or LDL cholesterol. Therefore, the inclusion of these studies would be expected to have bias these findings towards null.2 In a sub-analysis excluding the 6 studies that controlled for either serum or LDL cholesterol, saturated fat was associated with a statistically significant 30% increased risk of death from coronary heart disease (Fig. 2). Interestingly, in a sub-analysis including only the 6 studies which controlled for either serum or LDL cholesterol, saturated fat was associated with a statistically significant 18% increased risk of death from coronary heart disease (RR=1.18 [95% CI, 1.01-1.37]). This suggests that the adverse effects of saturated fat may extend beyond on simple measures of blood lipids.

FIGURE 2. Risk ratios and 95% CIs for fully adjusted random-effects models examining associations between saturated fat intake in relation to coronary heart disease mortality. SAT, saturated fat intake.

As coronary heart disease is the leading cause of death in the world, naturally these findings should be a cause for concern.28 Nevertheless, both the Siri-Tarino and Chowdhury meta-analyses are widely cited by proponents of Low-Carb and Paleo diets as providing compelling evidence in favor of a diet rich in saturated fat. It is important to note, however, that in the studies included in this meta-analysis, the difference for high vs low intake of saturated fat only ranged between about 5% and 10% of energy. This suggests that individuals following popular variants of these diets which often emphasize far higher intakes of saturated fat than recommended levels may be at a much greater risk of death.

It is important to note that the influence that saturated fat has on the risk of disease is not primarily determined by intake per se, but by which foods saturated fat is substituted for. As the intake of dietary fiber was universally low among subjects in these studies, this suggests that subjects consuming diets lower in saturated fat were substituting saturated fat primarily with lean animal foods and heavily processed foods.29 As dietary fiber was associated with a decreased risk of death from coronary heart disease in a number of these studies, this suggests that compared to fiber-rich foods, foods rich in saturated fat may be associated with an even stronger risk of coronary heart disease death.29

Although in this meta-analysis, the Israeli Ischemic Heart Disease Study appeared the least favorable of the hypothesis that saturated fat increases the risk of death from coronary heart disease, it should be noted that not only were the estimates controlled for serum cholesterol, in this study, saturated fat as a percentage of fat was actually associated with a statistically significant increased risk of death from coronary heart disease. In addition, subjects who were classified as being most adherent to religious Orthodoxy, which is typically accompanied by fasting periods in which the consumption of meat and other foods rich in saturated animal fat are prohibited, experienced a significantly lower death rate of coronary heart disease.11 This observation is supported by several other studies which found that Orthodox fasting is associated with improved cardiovascular risk factors, including blood lipids.30

The findings from this meta-analysis are in agreement with the demonstrated unequivocal causal relationship between diets rich in cholesterol and saturated fat, and the development of atherosclerosis in nonhuman primates, among dozens of other animal species. It had also been demonstrated that such diets cause heart attacks, and even cardiovascular related deaths in nonhuman primates at a rate similar to that of high-risk populations living in developed nations.31

The findings from this meta-analysis are also in agreement with numerous longitudinal ecological studies. For example, intake of saturated fat explained about 88% of the variance in death from coronary heart disease between the 16 cohorts in the 25-year follow-up of the Seven Countries Study.32 Similar estimates were also found for foods rich in saturated fat, including butter, meat, and animal foods combined.33 Similarly, in 1989, Epstein examined the changes in death from coronary heart disease in 27 countries during the previous 10 to 25 years, and noted that:
In almost all of the countries with major falls or rises in CHD mortality, there are, respectively, corresponding decreases or increases in animal fat consumption...
Epstein also noted that a number of other risk factors, such as smoking could not explain these findings, as the prevalence of smoking among women either remained largely unchanged or increased in most nations during this period, yet similar declines in death were often observed in both men and women.34 Epstein's findings are further supported by a number of studies that have incorporated the IMPACT CHD mortality model, which has been shown to adequately explain which risk factors and treatments that have contributed most significantly to the changes of rates of coronary heart disease mortality throughout most parts of the world.35


Dietary Patterns and Coronary Heart Disease Mortality


Dietary patterns characterized by high or low intakes of saturated fat may provide indirect evidence of the effect saturated fat has on the risk of death from coronary heart disease. For example, a recent meta-analysis of prospective cohort studies by Noto and colleagues found that both low-carbohydrate, and low-carbohydrate, high-protein diets, which highly correlated with saturated fat intake were associated with a statistically significant increased risk of death from all causes combined.36 Conversely, the findings for death from cardiovascular disease, although positive, were not statistically significant. However, several of the studies controlled for saturated intake, and sub-analyses in several of the studies found that the excess risk of death was greater when either saturated fat intake was above the median, or when the diets were classified as being animal based.37 38 Similarly, a sub-analysis in one of the studies found that the association with death was stronger after excluding nonadequate dietary reporters.39

These sub-analyses would likely allow for a clearer interpretation of the effects of a carbohydrate restricted diet rich in saturated fat. Unfortunately, estimates based on these sub-analyses were not provided in the meta-analysis by Noto and colleagues. Therefore, I performed a meta-analysis based on the studies included in the meta-analysis by Noto and colleagues, while using the estimates for the sub-analyses described above where possible.37 38 39 40 41 For the statistical analysis I used the same methods described by Noto and colleagues.36

In this meta-analysis, a high low-carbohydrate score was associated with a statistically significant 15% increased risk of death from cardiovascular disease, for which only 2 of 4 cohorts did not control for saturated fat intake (Fig. 3). On the other hand, a high low-carbohydrate, high-protein score was associated with a statistically significant 100% increased risk of death from cardiovascular disease, for which 3 of 4 studies did not control for saturated fat intake (Fig. 4). 

FIGURE 3. Risk ratios and 95% CIs for fully adjusted random-effects models examining associations between low-carbohydrate diets in relation to cardiovascular disease mortality. ¹Studies that included adjustments for saturated fat intake.

FIGURE 4. Risk ratios and 95% CIs for fully adjusted random-effects models examining associations between low-carbohydrate, high-protein diets in relation to cardiovascular disease mortality. ¹Studies that included adjustments for saturated fat intake. LCHP, low-carbohydrate, high-protein diet.

As coronary heart disease is the number one cause of cardiovascular death in the nations where these studies were carried out, this provides indirect evidence that diets rich in saturated fat, at least in the context of a carbohydrate restricted diet, increases the risk of coronary heart disease. Furthermore, the difference in intake of saturated fat between the low and high low-carbohydrate scores was generally smaller than the difference of intake between popular low-carbohydrate diets and recommended levels, suggesting that individuals who follow more extreme variants of these diets may be at an even greater risk of death. As reviewed previously, these findings may be explained, in part, by a number of adverse effects that carbohydrate restricted diets have been shown to exert on cardiovascular risk factors. For example, recent meta-analyses of randomized controlled trials have found that compared to diets rich in nutrient poor, low-fiber carbohydrates, carbohydrate restricted diets raise LDL cholesterol and impair flow-mediated dilatation.42

Findings from prospective cohort studies comparing vegetarians characterized by consuming moderately low saturated fat diets and health conscious omnivores may provide further indirect evidence of the adverse effects of saturated fat. I showed previously in a meta-analysis of 7 prospective cohort studies that compared to vegetarians, health conscious omnivores experienced a statistically highly significant 32% increased risk of death from coronary heart disease (Fig. 5).43 44 45 46 47

FIGURE 5. Risk ratios and 95% CIs for fully adjusted random-effects models examining associations between omnivorous diets in relation to coronary heart disease mortality.

It is important to note that the omnivores in these studies had a relatively low intake of meat, suggesting that individuals following popular meat based diets may be at a greater risk of death. This suggestion is supported by a recent meta-analysis of prospective cohort studies which found that an increment of 1 mg/day of heme iron, found only in animal tissue is associated with a 27% increased risk of coronary heart disease.48 As reviewed previously, in these studies, the degree of reduction in risk of death from coronary heart disease observed in vegetarians in these studies was generally in proportion to the expected reduced risk based on the differences in levels of total and non-HDL cholesterol. This suggests that these results may, in part, be explained by differences in intake of saturated fat.

The findings of a pooled-analysis of 11 prospective cohort studies by Jakobsen and colleagues suggested that replacing saturated fat with polyunsaturated fat, but not monounsaturated fat or carbohydrate, was associated with a significantly decreased risk of death from coronary heart disease.49 However, the researchers pointed out that these findings should be interpreted with caution, as the main dietary source of monounsaturated fat in these cohorts was animal fat, whereas the quality of carbohydrate was not considered. In this pooled-analysis, dietary fiber intake was controlled for, essentially removing a primary benefit of replacing foods rich in saturated fat with carbohydrate. In a different pooled-analysis including virtually the same studies, an increment of 10 g/day of dietary fiber was associated with a 27% decreased risk of death from coronary heart disease.23 This suggests that replacing saturated fat with the equivalent energy from fiber-rich carbohydrate would likely be associated with a significantly reduced risk of death from coronary heart disease. This suggestion is supported by a different meta-analysis which found that an increment of about 2 servings a day of whole grains was associated with a 22% decreased risk of death from cardiovascular disease.50 Interestingly, even Siri-Tarino and colleagues concluded in a more recent paper that saturated fat should be replaced with polyunsaturated fat and whole grains in order to reduce the risk of cardiovascular disease.7


Saturated Fat is a Major Problem


The findings reviewed here support the hypothesis that saturated fat increases the risk of coronary heart disease mortality. Furthermore, as reviewed previously, evidence also suggests that the hazardous effects of diets rich in saturated fat are also applicable to diets rich in organic, grass-fed animal foods. However, saturated fat is only one of a number of problems as far as chronic diseases are concerned. The effect that a particular food has on the risk of coronary heart disease cannot be fully explained by saturated fat content alone, but rather by multiple nutrients that likely operate together in a complex manner to modify the risk of disease. Therefore, it may be more appropriate to focus attention on recommending healthy dietary patterns that are naturally low in saturated fat, while rich in dietary fiber and other beneficial nutrients; primarily, minimally processed, plant-based diets. Such a focus may be more effective to help lower the intake of saturated fat, while simultaneously improving overall dietary quality compared to the more contemporary reductionist approach of focusing on modifying single nutrients.

In forthcoming parts of this review, I will examine both the effects of dietary and total circulating concentrations of saturated fat on the risk of total incidence of coronary heart disease. In addition, I will examine a number of other important limitations of the studies included in these meta-analysis that may have bias these findings towards null.2 3 4 5 6


Study acronyms: ATBC, Alpha-Tocopherol Beta Carotene Study; BLSA, Baltimore Longitudinal Study of Aging; EPIC-Greece, European Prospective Investigation into Cancer Greece; EUROASPIRE, European Action on Secondary and Primary Prevention through intervention to reduce events; FHS, Framingham Heart Study; HLS, Health and Lifestyle Survey; HPFS, Health Professionals' Follow-Up Study; IBDH, Ireland-Boston Diet Heart Study; IIHD, Israeli Ischemic Heart Disease Study; JACC, Japan Collaborative Cohort Study; LRC, Lipid Research Clinics; MALMO, Malmo Diet and Cancer Study; NHS, Nurses' Health Study; SHS, Strong Heart Study; SWLHC, Swedish Women’s Lifestyle and Health Cohort; ULSAM; Uppsala Longitudinal Study of Adult Men; VIP, Västerbotten Intervention Program; WES, Western Electric Study.


Please post any comments in the Discussion Thread.