A few days ago the American Heart Association (AHA) published a Presidential Advisory to address the growing confusion regarding the health effects of saturated fat intake. They noted that several meta-analyses have been published over the last few years that come to different conclusions, with some showing no harmful effects from saturated fat intake. They attribute the difference in results to the different replacement macronutrient in the trials: some replace saturated fat with carbohydrate and some with unsaturated fats. They believe that the trials replacing with unsaturated fats improved outcomes while trials that replaced with carbohydrate did not. They attribute this lack of benefit to increases in sugar and refined carbohydrate intake when carbohydrates replace saturated fat.
In this report, the AHA focuses on four randomized trials it considers to be the "core evidence". They claim that these trials were chosen because they represent the highest quality trials to base recommendations on. They claim that the trials compare high saturated fat to high polyunsaturated fat diets, did not include trans-fats as an intervention, controlled the intake of the intervention and control groups, had at least two years of intervention, measured biomarkers to assess adherence to the diets, and collected validated outcome data on CVD events. Other trials were acknowledged as "noncore evidence" because they had one or more flaws that limited their value.
The AHA continued this report by reviewing the four core trials and completing a meta-analysis of their own. I was so shocked and disappointed by the incorrect interpretations, misrepresentation of trials, and cherry-picking within this report that I had to write a review. I have previously reviewed the four core trials and published detailed analyses in the Schwanz Nutrition Journal, so it was very easy to spot the errors and misrepresentations made by the AHA. Two of these trials were included in the first two issues of my journal which were released for free, so I will go into more detail on those. I feel it would be unfair to paying subscribers to go into detail on the other two trials (LA Veterans Trial and the Finnish Mental Hospital Study), so only highlights will be included here.
One of the core trials was covered by the AHA with only a short paragraph. This was the British Medical Research Council's study on soybean oil. This trial primarily replaced saturated fat intake with soybean oil. It likely received only a brief mention from the AHA because it showed no differences in outcomes between the two groups. It was a fairly well done study that produced a large difference in saturated fat and polyunsaturated fat intake between the two groups and had about half of the study population maintain their diets for over four years. Interestingly enough, of the four trials cited as the core evidence, this trial reduced cholesterol levels in the intervention group the most.
The Oslo Diet-Heart trial was another core trial that was butchered by the AHA to the extent that it would make a perfect assignment for students to try to identify as many glaring omissions, errors, and misrepresentations as they can. This was represented by the AHA as a trial that had a control group with a high saturated fat intake and an intervention group that decreased saturated fat and increased polyunsaturated fat intake. This led to decreases in cholesterol levels and cardiovascular outcomes in the intervention group. The AHA gives the impression that changes in saturated fat and polyunsaturated fat intake were the only dietary changes made. In reality, there were many other recommendations that were made to the intervention group which included recommendations to increase consumption of fish, salads, fruit, nuts, peas, carrots, and beans while reducing intake of sugar. Margarine intake was another significant confounder that received no mention from the AHA. Average margarine consumption in Norway during the study period was estimated to be 65 grams per day. Additionally, an estimated 70% of the fat in margarines at this time in Norway came from hydrogenated oils. These margarines were "entirely restricted" in the intervention group, while the controls continued consumption. Subjects in the intervention group also were provided considerable amounts of sardines canned in cod liver oil, which supplied a significant amount of naturally occurring vitamins A and D. If all of these confounders (none of which were mentioned by the AHA) weren't enough, consider this: during the time of this trial, Norway was experiencing a drastic increase in mortality from myocardial infarction, which occurred almost entirely in men. This rate started to decline after this study was completed. Thus, the results of this trial are most likely not relevant to other populations. The AHA made one last blatant misrepresentation of the data by stating that the intervention group continued to experience reduced cardiovascular mortality for an additional six years after the trial ended. At the end of the trial there was no statistically significant difference in cardiovascular or total mortality between the groups, though the intervention group did have a lower number. At the 11-year mark that the AHA refers to, the cardiovascular mortality is still no different between the groups (88 deaths in intervention group vs 102 deaths in controls; P=0.13); total mortality also showed no difference (101 deaths in the intervention group vs 108 in controls).
The other two trials will be covered in less detail, only covering some of the highlights of the errors made by the AHA. The first trial was the LA Veterans Trial. The AHA reports that the "average duration" was eight years of intervention. Though unclear, this seems to refer to the length of time that subjects followed the intervention diets. I was unable to find an average length of time on the diets in the original study publication, but the trial was completed over the course of about eight years, with many subjects entering the study after the start of the trial, making it very unlikely that the average duration was eight years. In a publication from the Cochrane Collaboration(1) that reviewed this trial, it reported an average of 3.7 years following the diets. The AHA commended the "strict dietary control" of the trial, possibly because the meals were provided for the subjects. However, they neglected to mention how many of their meals the subjects actually consumed at the center providing the meals. In the intervention group, less than half of their meals during the study period were eaten at the center. Controls ate 56% of their meals at the center. The AHA refers to this study as a high quality, well-controlled trial that used vegetable oils to replace saturated fats in the diet. I'll pull a quote directly from the original publication to make it more clear that this intervention was not just a reduction in saturated fat intake: "This restriction led to a decision to test a replacement of the saturated animal fats and hydrogenated shortenings of the conventional diet by equal quantities of unsaturated fat in the form of vegetable oils in the experimental diet." There was NO mention of hydrogenated shortenings being reduced in this trial in the AHA advisory. Let's not forget what the AHA set as one of the exclusion criteria to be a core trial: trans-fats were not a major component of the intervention.
The last trial was the Finnish Mental Hospital Study. Others have commented before (and I completely agree) that this trial is a perfect litmus test for meta-analyses or review articles related to saturated fat. Any review or meta-analysis that includes this trial as a legitimate and meaningful trial should not be taken seriously. This trial has numerous critical flaws in addition to many smaller flaws. It wasn't really a randomized trial, it wasn't a controlled trial, it didn't come close to isolating saturated fat for polyunsaturated fat. A few of the differences in intake other than SFA/PUFA included sugar, ice cream, hard margarines, and cardiotoxic medications. How many of these differences were mentioned in the AHA report? If you guessed zero, you are correct! Another interesting point: this is a cross-over trial. When the outcomes of a trial are long-term chronic disease events, a crossover design is completely inappropriate. One last criticism is the reporting of outcome data from the AHA: the outcomes in this report are different from what I found, and reported as much more favorable by the AHA. Quite frankly, it's not worth the time to dig through the numerous publications on this study to verify those numbers.
The AHA next completed a meta-analysis by combining the four core trials. This can be fully summarized in four words (or should it be three?): garbage in, garbage out. Instead of looking at a specific outcome, such as total mortality, cardiovascular mortality, or myocardial infarctions, they chose to use the primary outcome from each trial, even though they may be different, and combine them. They chose this method because a posthoc analysis that picks an outcome such as cardiovascular mortality after the trials are completed will be subject to more bias compared to a prospectively chosen outcome. This is true, but here is the problem: the primary outcome from each trial is most likely a posthoc outcome anyway. These trials were all completed prior to the popular use of clinical trial registers where researchers declare their primary outcomes before completing the trial. Clinical trial registers became popular because older trials, such as the "core four" would commonly measure numerous variables and cherry-pick the most favorable outcomes as the "primary" measure. Also, not every trial will choose only one primary outcome. How the AHA determined the one primary outcome to include from each trial was not reported, nor was the actual measurement used. In the meta-analysis, the trial that showed the most favorable results (in terms of better outcomes for reduced SFA intake) was the Finnish Mental Hospital study; this trial also carried more "weight" in the meta-analysis than any of the other three trials because it had more outcomes. Here is the most fatal flaw: the AHA used results from "the entire Finnish trial population" which included subjects with existing heart disease as well as individuals free of heart disease at baseline. If the AHA used the primary outcome as they claimed, they would NOT include the entire population. Doing this significantly increases the amount of subjects and the weight of this trial in the meta-analysis. Here is a quote from one of the many original publications on the Finnish trial that makes it very clear which population was the primary focus: "In principle all patients within the specified age range were included in the study. As, however, this was meant to be a trial of primary prevention, those patients who at the outset showed evidence of CHD were excluded(2)." Based on their own exclusion criteria to qualify as a core trial, at least 3/4 studies presented are clearly inappropriate. The Medical Research Council Soybean trial was the only trial coming close to meeting the criteria defined by the AHA. Excluded trials were also described (with similar quality as the first four) by the AHA.
I hope this review opens the eyes of anybody that still considers the American Heart Association to be an authority on nutrition. A quote comes to mind that perfectly reflects my view on publications coming from the AHA: "The greatest enemy of knowledge is not ignorance, it is the illusion of knowledge." No publications would be much better than highly flawed publications. It would be in the public's best interest if the American Heart Association would stop publishing on the topic of nutrition.
1. Hoooper, L. et al. Reduced or modified dietary fat for preventing cardiovascular disease. Cochrane Database of Systematic Reviews 2012. Issue 5.
2. Miettinen, M et al. Dietary prevention of coronary heart disease in women: the Finnish Mental Hospital study. International Journal of Epidemiology 1983. 12(1):17-25.