by Tim Noakes
4
The Centenary Debate
‘If you must criticize scholars whose work challenges yours, do so on the evidence, not by poisoning the land on which we all live.’
– Alice Dreger, Galileo’s Middle Finger1
In November 2012, I hosted Swedish physician Dr Andreas Eenfeldt and health coach Monique Forslund in Cape Town to launch the English version of Forslund’s book Low-carb Living For Families. We used the occasion to have both speak at an evening seminar at the UCT Faculty of Health Sciences. The audience was riveted. Eenfeldt has been a key figure in the global LCHF movement since he began to promote the LCHF lifestyle in his medical practice in Karlstad, Sweden. He currently hosts the world’s most widely accessed, well-researched and well-referenced LCHF website, Diet Doctor.
Over dinner one evening, I happened to mention that UCT had invited me to debate Professor Jacques Rossouw in its Faculty of Health Sciences Centenary Debate in a few days’ time, on 6 December 2012. Eenfeldt immediately asked: ‘Do you know that the most important finding from his paper on the Women’s Health Initiative is a single sentence hidden deep in the text and not repeated in the abstract?’
I did not know. Eenfeldt pulled out his computer and showed me the evidence. There, on the seventh page of a 12-page article in a 2006 issue of JAMA was the key line: ‘The HR [hazard ratio] for the 3.4% of women with CVD at baseline was 1.26 (95% CI [confidence interval], 1.03-1.54).’2
This means that the women in the trial who had suffered a previous heart attack and who were randomly selected to eat the ‘heart-healthy’ low-fat diet for the duration of the trial were at a 26 per cent increased (relative) risk of experiencing further heart problems during the eight years of the trial compared with those on the supposedly unhealthy, artery-clogging, higher-fat diet. Although this was the sole significant finding of the diet component of a $700-million study, no one paid any attention. Instead, the authors simply buried it, out of sight, out of mind.
Next, Eenfeldt drew my attention to Figure 3 on the ninth page. ‘See here,’ he said, ‘there’s a phrase missing. It’s the line identifying the increased risk for women with previous heart disease.’ Thus, if you missed the key sentence buried in the article and only looked at this table, you would never know about the study’s only significant finding.
Another paper, published in 2011, five years after the original publication, reported that within as little as a year, the condition of those women who started the trial with T2DM and who were randomised to the low-fat diet had worsened. The authors were sufficiently transparent to acknowledge this, for they wrote that ‘women with diabetes at baseline did experience adverse glycemic effects of the low-fat diet, which indicated that caution should be exercised in recommending a reduction in overall dietary fat in women with diabetes unless accompanied by additional recommendations to guide carbohydrate intake’.3
However, they should have said that their data showed that patients with T2DM should not be treated with a low-fat diet. These adverse effects were reported after only the first year of an eight-year trial. To my knowledge, the final eight-year outcome data for women with T2DM has never been published. If diabetic outcomes worsened after only one year on a low-fat diet, how much worse might those outcomes have been seven years later?
If the Women’s Health Initiative trial had been properly designed to detect this adverse outcome, I wondered whether the researchers should have taken women with T2DM off the low-fat diet after the first year of the study. Those women would have presumed that their health would not be allowed to deteriorate as a result of participation in an ethical trial. Of course, terminating that part of the trial would have had major consequences for those who promote the low-fat diet dogma.
Convinced that exposing this new information about the concealed aspects of the Women’s Health Initiative study, which Rossouw had directed, would swing the Centenary Debate decisively in my favour, I set about finalising my presentation. I was certain that my UCT colleagues would want to hear the truth.
Clearly, I was still unaware of the power of the omertà.
I have known Rossouw since the 1980s. In those years, I was a peripheral member of a research team under his direction that completed a population intervention trial in three towns in the Western Cape, the so-called Coronary Risk Factor Intervention Study (CORIS). The goal was to determine whether an intensive, hands-on, ‘heart-healthy’ intervention programme in one town, Robertson, would produce better long-term health outcomes than a less intensive intervention in a second town, Riversdal.
In Riversdal, the sole source of ‘heart-healthy’ information was what the project organisers sent through the mail. Robertson, the intensive, hands-on town, received the same ‘heart-healthy’ information focusing on a low-fat diet and physical activity, but also specific interventions to improve blood pressure control, and reduce stress and blood cholesterol concentrations, especially in those at highest risk. A third town, Swellendam, served as the control. Swellendam received only the general health information available to all South Africans at that time through radio, television and the printed press.
Four years later, the researchers returned to measure the effects of the intervention in the different towns. Surprisingly, they found the outcomes were essentially the same in the two intervention towns. In other words, the more intensive programme targeting high-risk individuals in Robertson had not produced any additional benefits than those measured in Riversdal.4
Eight years later, a second follow-up found that although the risk-factor profiles had continued to improve in all three towns, the extent of improvements in the low-intervention (Riversdal) and control (Swellendam) towns were greater than in Robertson, the town receiving the most intensive intervention. The authors’ final conclusion was that the low-intensity intervention was the most successful.5
So, according to the scientifically revered null hypothesis,1 this study actually proved that a ‘heart-healthy’, hands-on intervention was no better than simply giving general health advice through the mail, or letting participants educate themselves on the basis of what they hear or read in the media. But how could this possibly be true if the intervention in Robertson was based on the best available, evidence-based medical practices?
An inconvenient conclusion would have been that the ‘best’ evidence-based medical practices available in 1980 caused harm. However, this was not how Rossouw and his colleagues – including, by association, me – presented the study outcomes. Instead, they concluded that such intervention trials can no longer measure the ‘truth’, positing a new post-hoc theory:2 that already by the early 1980s, all the dietary and other information necessary to improve a community’s health was freely available in the popular media. Thus, providing yet more information, even accompanying an intensive, hands-on approach to those at greatest risk, could not produce a better outcome.
However, objective science does not allow such post-hoc interpretations. That possibility should have been included in the original hypothesis. The original hypothesis was whether or not the intensive intervention programme would improve health outcomes more than doing little or nothing in Riversdal and Swellendam. It did not. According to the null hypothesis, the only scientifically legitimate conclusion was that the intensive intervention – including the promotion of the ‘heart-healthy’ low-fat diet – failed to make any difference, and might even have caused harm. To admit this conclusion would be to break the omertà. Rather than tell the truth, the researchers chose to invoke the code of silentce.
Importantly, a key focus of the intervention was the promotion of a ‘heart-healthy’ low-fat diet to lower blood cholesterol concentrations. In these farming communities, which would, at least in the past, have eaten a higher-fat (although not necessarily a lower-carbohydrate) diet, this might have meant quite a large change from their more traditional farmers’ diet.
What if the ‘heart-healthy’ diet had been adopted more widely in Robertson than in the other
two towns? And what if this diet produced an adverse outcome that outweighed other health benefits produced by the other interventions promoted exclusively in that town? Interestingly, the diet had little effect on weight in any of the towns, as weight increased over the 12 years in men, but fell marginally in women.6 Without a change in body mass towards more healthy values, it would be unlikely that long-term health would change significantly in a community at risk from a whole range of IR conditions.
Perhaps the key conclusion from this study was that the ‘heart-healthy’ low-fat diet had failed to improve the health of the people living in Robertson. Had that lesson been learnt, it might have influenced the planning of the most expensive diet trial ever undertaken, the $700-million Women’s Health Initiative Randomized Controlled Dietary Modification Trial (WHIRCDMT) in the US, funded by the National Institutes of Health. It was a study with which Rossouw would soon become intimately involved.
In 1989, shortly after completing the initial phase of the CORIS, Rossouw immigrated to the US and in time became an American citizen. In 1991, he was placed in charge of the WHIRCDMT, an eight-year RCT involving more than 48 000 post-menopausal women. It included elements of the CORIS writ very, very large.
Rossouw’s involvement in the study would initially do no harm to his international reputation. In 2006, Time magazine named him one of its 100 most influential persons in the world. A special event to acknowledge this achievement was held at UCT in May 2006. His son, Jacques Rousseau, represented him. Rousseau said of his father: ‘What drives him is doing his bit to decrease ignorance and confusion, and he’ll continue to hold that committed attitude for as long as he’s got something useful to say – regardless of whether it’s appreciated or not.’
I suspect that the NIH funded the WHIRCDMT specifically to prove that the low-fat diet was the perfect eating plan for all humans. It was an attempt by the US government, through its medical research agency, the NIH, to prove once and for all that the 1977 United States Department of Agriculture (USDA) Dietary Guidelines for Americans, which promote the consumption of cereals and grains, seed (vegetable) oils and, in practice but not by design, high-fructose corn syrup (HFCS), were fully justified.
Enjoying such an unprecedented level of institutional support, senior researchers involved in the WHIRCDMT, including Rossouw, might reasonably have begun to dream that perhaps even a Nobel Prize was within their grasp.
This then was the man that the UCT Faculty of Health Sciences had lined up as my opponent in the Centenary Debate. They may have assumed that it would not be a close contest; the WHIRCDMT Goliath would not fall to this David, a mere ‘sports scientist/general practitioner’.
The topic for the debate was ‘Cholesterol is not an important risk factor for heart disease and current dietary recommendations do more harm than good’. I had no input in the choice of topic. My interest is the role of high-carbohydrate diets in causing a constellation of chronic diseases in people with IR, and the reversal of those conditions with carbohydrate restriction.
I opened the debate in a packed lecture theatre. I had recently returned from a lecture trip to the Eastern Cape, where two university audiences had visually and verbally expressed their enjoyment of my lectures. On my home turf, in front of my own faculty, I expected a similar response, even though I knew it was not going to be an easy encounter.
The first surprise came in the week before the debate, when UCT’s chosen moderator, Professor Jimmy Volmink, dean of the Faculty of Medicine and Health Sciences at the University of Stellenbosch, asked for a copy of my slides and notes. He needed them, he told me, to prepare his closing remarks. Volmink is also director of the South African Cochrane Centre at the South African Medical Research Council (SAMRC).
At the time, I was unaware that his institute had already been commissioned, ostensibly by the Heart and Stroke Foundation of South Africa (HSFSA), to prepare a meta-analysis of studies of low-carbohydrate diets, of which Volmink turned out to be the senior co-author. The clear goal of that meta-analysis, as I discuss in Chapter 6, was to prove that LCHF diets are of no medical value and are likely very dangerous.
Therefore, already in December 2012, Volmink was not an independent party to this debate. Published in 2014, the study would become known as the Naudé review, named after its first author, Dr Celeste Naudé.7 The study would play a pivotal role in the HPCSA’s decision to charge me with unprofessional conduct. Unfortunately, the study contained many significant errors which, when corrected, reversed its conclusions (see Chapter 6).
When I told my wife, Marilyn, about Volmink’s request, she was concerned. But how could I not trust Volmink, dean of medicine at such a prestigious university? If you cannot trust a dean who is also a director at the SAMRC, then whom can you trust in the hallowed research halls?
In my lecture, I decided to cover five separate topics:8
Economic considerations drove the adoption of the current dietary guidelines without proper scientific evaluation or proof.
Within five years of widespread adoption of these guidelines, rates of diabetes and obesity increased explosively.
The presence of the genetic predisposing condition known as insulin resistance explains why large numbers of people in predisposed populations become obese and diabetic when exposed to a high-carbohydrate diet.
A high-fat diet reverses all known coronary risk factors in people with IR, whereas a high-carbohydrate diet worsens those factors.
The 48 836-person Woman’s Health Initiative study, of which Rossouw was project director, proves that the 1977 USDA dietary guidelines accelerate disease progression in people with either known heart disease or diabetes. Thus, this landmark study provides definitive evidence disproving the diet-heart hypothesis, which Dr Ancel Keys promoted in the US beginning in the 1950s.
For the first 28 minutes of the lecture, I methodically repeated key pieces of evidence on the first four topics as I had previously presented to the faculty in July and in presentations with Eenfeldt and Forslund. This was evidence that, in my opinion, proved that the diet-heart and lipid hypotheses are just hypotheses. Worse, they are mythical, with no substance in fact.
In my judgement, the killer information that would decide the debate was contained in my fifth point. I would argue that the findings of the $700-million WHIRCDMT were definitive proof that the low-fat diet does not reduce the risk of developing heart disease, T2DM or cancer, and that the low-fat diet has no effect in the prevention or reversal of obesity.
The WHIRCDMT had shown that the health of those who were the sickest, because they already had heart disease when the study began, worsened more rapidly if they were placed on the low-fat diet. This is a classic ‘canary in the coal mine’ finding, for if a diet is dangerous, who will be the first to provide evidence of that ‘danger’? Clearly, those who are at the greatest risk, because they already have underlying pathology. In contrast, those who are initially the healthiest may not show any detrimental effects of a bad diet if the trial is too short for the effects to become apparent. We call this a false negative finding; because of a flaw in its experimental design, a study fails to detect a detrimental outcome that is real.
So this iconic study designed to prove the health benefits of the low-fat diet had, in fact, proved the opposite. And the WHIRCDMT is not the sole RCT to have found this. Two additional studies, which have cost US taxpayers hundreds of millions of dollars more, confirmed this finding.9
I was certain that after I had presented this evidence, I would be judged the ‘winner’ of the debate. Surely, I thought, an unbiased UCT audience would recognise that Rossouw himself had provided evidence that his theory was wrong; and that continuing to defend a theory that his own work had disproved was neither academically nor ethically viable.
As I was about to introduce my killer blow, I got the second surprise. Volmink said that my time was up and I should stop talking. I told him that was censorship and simply carried on. Interestingly, a YouTube video of the
debate shows that Volmink interrupted me 28 minutes into my presentation, before the 30-minute time limit was up. Later, I learnt that a cadre of medical professors sitting near the front of the lecture theatre was busily exchanging notes that were eventually passed on to Volmink. Were the professors trying to stop me before I reached my fifth point? Was it their decision that Volmink should use the excuse that I had exceeded my time limit?
I hurriedly presented the final, critical pieces of evidence, which included exposing the missing line in Figure 3 of the original WHIRCDMT publication. (Today, more than a decade after its publication, that error has yet to be corrected, even though I drew attention to it in a scientific article in 2013.10 Rossouw responded,11 but still failed to address the substantive issues that I had raised.12)
I ended my lecture by showing the reluctance of Rossouw and his employer, the NIH, to acknowledge the gravity of their negative finding and to admit that, according to the null hypothesis, the WHIRCDMT disproved the diet-heart hypothesis.3 I included reference to what was said at the NIH press conference to announce the (disappointing) results of the trial.
There, cardiologist Dr Elizabeth Nabel, the director of the funding body, the National Heart, Lung, and Blood Institute, had said: ‘The results of this study do not change established recommendations on disease prevention. Women should continue to … work with their doctors to reduce their risks for heart disease including following a diet low in saturated fat, trans fat and cholesterol.’
Her employee, Rossouw, added: ‘This study shows that just reducing total fat intake does not go far enough to have an impact on heart disease risk. While the participants’ overall change in LDL “bad” cholesterol was small, we saw trends towards greater reductions in cholesterol and heart disease risk in women eating less saturated and trans fat.’
In my judgement, I had exposed the key paradox in the Centenary Debate: UCT had chosen the one scientist in the world who had participated in two studies costing hundreds of millions of dollars, and which proved that the low-fat diet is without long-term health benefits and could, in fact, be harmful, to present the opposite argument. How had an institution publicly committed to academic excellence and the search for truth allowed this to happen?