Sunday, April 26, 2015

Testing the Diet Heart Hypothesis in Women

Most of the diet heart trials were only conducted in men probably because CHD is thought to be the ‘fat, middle-aged man’s disease’, but CHD is the leading cause of death in Australian [1] and American [2] women as well.  In Australia, 45.5% of all deaths from CHD occur in women [1].  So while CHD certainly does affect more men than women, this difference seems to be greatly exaggerated.

Despite these statistics, only two of the trials included women (Finnish Mental Hospital Study (FMHS) and Minnesota Coronary Survey (MCS)) and consequently the weighting of events/deaths for women is far below 45.5%

Weighting in ‘Adequately Controlled Trials’
Weighting in All Trials
Men (%)
Women (%)
Men (%)
Women (%)
Major CHD Events (E)
Major CHD Events (P)
Total CHD Events (E)
Total CHD Events (P)
CHD Mortality
Total Mortality

This goes to larger issue in medicine where women are often underrepresented in clinical trials, sometimes even to the extent where women are prescribed a drug which hasn’t even been tested in a single woman in a clinical trial [3].  A lot of small, short term studies are done only using men, as the menstrual cycle may affect the results.  While this is a legitimate concern of small, short term studies, it would have little to no impact in larger, long term studies such as the diet heart trials, particularly if they included mainly postmenopausal women

This is important because certain interventions can affect women differently.  There are a few examples in [3] and [4] and plenty on Suppversity if you do a few searches.  Another example may be statins, as a meta-analysis found that in secondary prevention trials statins don’t significantly reduce total mortality in women, whereas they do with men [5].

Replacing SFA with PUFA may be another example where women respond differently as there was a fairly consistent pattern in MCS and FMHS: when the RRs for men were compared with those for women: men do better than women or women do worse than men (depending on how you look at it).

RR for Men
RR for Women
Major CHD Events
Total CHD Events
CHD Mortality
Total Mortality

RR for Men
RR for Women
Major CHD Events
CHD Mortality
Total Mortality

Given that women were very underrepresented in the diet heart trials and respond to replacing SFA with PUFA less favourably/worse than men do, I wonder what the results would be if more of the trials included women:

  • Would this pattern persist?  Is this a real effect?
  • Would the pooled result be completely neutral even without excluding inadequately randomised and/or inadequately controlled trials?
  • If there were some repeats of MCS, would health authorities be urging women away from vegetable oils?

The answers to those questions is debatable, simply because we don’t know and I don’t think we can draw too strong a conclusion from two trials (especially because one of them (FMHS) was both inadequately randomised and inadequately controlled).  This is an additional reason to conduct another trial where SFA is replaced with PUFA, one which is randomised, well controlled and includes both men and women.  Conventional dietitians/doctors should be motivated to do so, not only do they have the burden of proof, but also because the current results for women aren't too favourable.

Monday, April 20, 2015

Does Reducing Saturated Fat Lower the Risk of Coronary Heart Disease?

Had you asked this question perhaps only 10 years ago the only response you may have got would be ‘yes’.  If you asked the same question today, an increasing number of people would answer ‘it depends what you replace it with’, but I would answer ‘most likely not’.  To understand these points of view we need to begin with a little history (don’t worry I won’t bore you with Ancel Keys)

The Lipid Hypothesis and the Diet Heart Hypothesis

The lipid hypothesis was based on the results of the Framingham study, that serum cholesterol levels were associated with the incidence of coronary heart disease (CHD).  As a result, the hypothesis was that “measures used to lower the plasma lipids in patients with hyperlipidemia will lead to reductions in new events of coronary heart disease” [1]

When the lipid hypothesis was originally developed they couldn’t differentiate between the cholesterol contained in different lipoproteins (e.g. HDL-C and LDL-C), so all mentions to cholesterol in the blood referred to total cholesterol (total-C).  Metabolic studies conducted at this time found that SFA increased total-C, PUFA decreased total-C and MUFA had no noticeable effect [2], which is consistent with current research [3].  Whether dietary cholesterol increased total-C was still debatable, though many suspected it did, a view that was supported by later research in the 60’s and 70’s [4].

The lipid hypothesis combined with these metabolic studies led to the diet heart hypothesis, which is that by reducing total-C, reducing SFA or replacing SFA with PUFA would be expected to reduce the incidence of CHD.  The diet heart hypothesis was tested in a number of clinical trials, most of which were conducted during the 60’s and 70’s, that sought to aggressively lower total-C through a substantial reductions in SFA and dietary cholesterol and doubling or trebling PUFA intake.  Very few trials used a low/reduced fat diet and when it was used or recommended (e.g. by the AHA or dietary guidelines) reducing one’s fat intake was simply seen a means to lower SFA intake [5], which increased total-C by about twice as much as PUFA decreased it [6]

Low Fat for Coronary Heart Disease

Low fat diets became popular for weight loss and prevention of CHD around the 80’s and 90’s.  However, the incidence of obesity increased in many industrialised countries during this period, due to an increase in energy intake (without a parallel increase in energy expenditure).  In the US this was due to an increase in carbohydrates.  As a result, the percentage of calories from fat declined, but absolute fat intake stayed the same [7].  The relationship between low fat diet advice and the obesity epidemic would later lead to some fairly widespread scepticism about low fat diets, which may have been part of the reason why several health organisations later retracted their advice to eat a low fat diet.

Low fat diets declined in popularity in the 2000’s.  In the later 20th century people were able to differentiate between HDL-C and LDL-C, etc.  HDL-C was found to be inversely associated with CHD and the total-C:HDL-C ratio emerged as a far more predictive risk factor than total-C, taking into account that inverse association [8].  Replacing SFA with carbohydrates was found to not improve the total-C:HDL-C ratio [3].  In addition, there were concerns that increasing carbohydrates would elevate triglycerides and decrease LDL particle size [9] [10], altogether making carbohydrates an unattractive replacement for SFA.  In support of their effects on the total-C:HDL-C ratio, meta-analyses of observational studies were not finding that total fat was associated with CHD [11] [12] and/or that carbohydrate or replacing SFA with carbohydrate was inversely associated with CHD [13] [14]

At about the same time, interest in low carbohydrate diets was reignited and a number of clinical trials compared a low fat diet to a low carb diet.  In these trials the low carb diet was equal to or better than the low fat diet regarding weight loss and blood lipids [15].  The improvement in blood lipids is what you would expect from replacing carbohydrate with fat [3].  Meanwhile the low fat diet Women’s Health Initiative (which decreased fat intake by 9% of total calories, while increasing fruit and vegetable consumption) pretty much accomplished nothing besides affirming the null hypothesis [16]

The Return to Fat Modification

As a result, many health organisations currently say that they no longer recommend a low fat diet (which depends on how you define low fat [17]) and recommend replacing SFA with unsaturated fats, rather than carbohydrates.  This is largely because replacing SFA with either MUFA or PUFA would be expected to lower LDL-C (which some people still use) and the total-C:HDL-C ratio (as opposed to the days of total-C where MUFA was ignored because it had no noticeable effect) [3].  Some will distinguish between MUFA and PUFA, as meta-analyses of observational studies have not shown replacing SFA with MUFA to reduce the risk of CHD [13] and also there haven’t been many clinical trials on CHD where SFA is replaced with MUFA.

Instead, these people recommend replacing SFA with PUFA, particularly linoleic acid (18:2 n-6).  In additional to how SFA and PUFA affect blood lipids, they may also cite evidence from observational studies and randomised controlled trials to support their position, often selecting the meta-analyses that support their position most strongly (e.g. citing [13] [18] [19] instead of [20] [21] [22] or the Cochrane meta-analysis [23])

Indeed, while advice to replace SFA with PUFA is very mainstream, there are some researchers who disagree with it.  This alternative position is often be argued from a different interpretation of the evidence [22] and proposed mechanisms suggesting that increasing omega 6 (or the omega 6 to omega 3 ratio) may be harmful [24]

My Perspective

With these conflicting views I decided to look at the clinical trials that replaced SFA with PUFA [25] and the major meta-analyses [26].  I found that sometimes the meta-analyses were inconsistent (the trials they included and which figures they used) and sometimes they made errors (some minor, some major) [27] [28] [29].  In addition, while looking at the trials it became clear very quickly that they generally weren’t well controlled interventions, almost always in favour of the experimental group (high PUFA group):

·         The experimental group were also advised to reduce their intake of TFA and/or foods rich in TFA such as baked goods [22]
·         In some trials the experimental diet was highly multifactorial, such as also increasing intake of fish and whole plant foods [30] [31]
·         Or there were other differences such as a lower and insufficient intake of vitamin E in the control group [32] or an increase in the use of cardiotoxic medications in the control group [33].

Therefore, I categorised trials as being ‘adequately controlled’ or ‘inadequately controlled’ and used software from the Cochrane Collaboration [34] to conduct my own informal meta-analysis, paying special attention to the pooled results of the ‘adequately controlled’ trials [35], as these most closely reflect a true test of replacing SFA with PUFA.  What I found was that there was no benefit (or hint of benefit) for replacing SFA with PUFA, the risk ratio for CHD events, CHD mortality and total mortality was essentially 1.0 [35].  This result was after excluding a very unfavourable trial that has been criticised for experimental group potentially having a higher intake of TFA (when stating their position these critics ignore that in the vast majority of the trials it’s the control group that has a higher intake of TFA, as well as all the other differences).

Therefore, my interpretation of the evidence is that:

·         Replacing SFA with carbohydrate, MUFA or PUFA is unlikely to reduce CHD and mortality

·         It’s unlikely that SFA is bad/toxic, or else the evidence suggests carbohydrate, MUFA and now PUFA would be equally as bad/toxic

Wednesday, April 15, 2015

Looking Back on the Latest AHSNZ Conference, and Moving Forward to the Next One

Surviving the Urban Jungle

A couple of weeks ago I had the pleasure of attending a conference in Wellington organised by the Ancestral Health Society of New Zealand (AHSNZ) called ‘Surviving the Urban Jungle’.  See the programme below:

Survival of the Fittest: Lessons from the Christchurch Earthquakes
Jamie Scott, Health Researcher, Synergy Health, Christchurch

Whakapapa or fatness? Assessing physical health and performance with measures that matter
Dr Isaac Warbrick, PhD, Auckland University of Technology, Auckland

Our love affair with speed: Why it really isn’t sexy
Lulu Loya Wu, Health Coach, Wellington

Fertility vs Famine – A delicate balance of survival
Kate Callaghan, Nutritionist, Wanaka

Surviving modernity’s ‘fadism': Exploring the ethics of ancestral health
Andrew Dickson, PhD, Massey University, Palmerston North

Weight loss: The pot of gold at the end of the rainbow
Dr Anastasia Boulais, Medical Practitioner, Christchurch

Max Bell
Shelley Bell
Jason Young

In particular, Jamie’s and Andrew’s talks really struck a chord with me.  Jamie’s talk was powerful and his message of being physically and mentally prepared for a natural disaster (or similar situations) is one that is quite unique.  Andrew’s talk offered insights on the pillars of modern faddism and methods to resist faddism, which is particularly relevant given the current state of Paleo

The movement workshop was fun and featured some basic parkour and mobility (don't be put off by the dramatic youtube videos of parkour, it's actually very scalable)

* The society has also previously organised two other conferences: the first in Christchurch called ‘Ancient Genes vs. Modern World’; and the second in Wanaka called ‘Health of the Land, Health of the People

Looking Back, Moving Forward

The society is also holding an international symposium in October at Queenstown called “Looking Back, Moving Forward”.  Among the impressive speakers list, made up a large number of professionals and academics, you’ll find many well-known people in the Paleo/Ancestral Health community, but also many people who aren’t.  A broad range of topics will be covered, such as diet, physical activity, urban design, indigenous health, sustainability and climate change.  This quote from the Whole 9 South Pacific facebook page sums up my thoughts about the event as well:

“Far from choir-preaching, this event has the ability to bring together many professionals whose work has never been appreciated by the ancestral community before. And, of course, there will be some of the very well known names. Check out the speaker line up”

Jamie was kind enough to invite me (a relative nobody) to be one of the speakers.  My talk will be based on my research on saturated fat and coronary heart disease (which I’m planning to publish as a meta-analysis around the middle of the year). 

You can register here.  There’s also information on how to get there, where to stay and what else you can do around Queenstown (if you can, turn the trip into a holiday, Queenstown is a great place for that)

Sunday, April 5, 2015

Paleolithic Diet Trials: Masharani, et al (2015)

Participants and Diets

25 adults aged 50-69, with type 2 diabetes, were randomised to follow a Paleo diet or the American Diabetes Association diet.  Three meals and three snacks were provided.  An attempt to avoid weight loss was made by calculating maintenance energy needs and then adjusting calorie intake if 3lb (~1.35kg) of weight loss occurred.

“The Paleo diet consisted of meat, fish, poultry, eggs, fruit, vegetables, tree nuts, canola oil, mayonnaise and honey. We excluded dairy products, legumes, cereals, grains, potatoes and products containing potassium chloride. Some foods, such as mayonnaise, carrot juice and domestic meat, were not consumed by hunter gatherers but contain the general nutritional characteristics of pre-agricultural foods.”

In addition, there were 3 ramp up diets for 7 days for those in the Paleo group to “allow for adaptation of the subjects’ intestinal tract and potassium handling systems to adjust to the markedly higher dietary content of the fiber and potassium in the Paleo diet”

The ADA diet is not defined in the paper except for “…containing moderate salt intake, low-fat dairy, whole grains and legumes”, as opposed to the exclusion of these foods in the Paleo diet (excluding added salt, not all salt).  There is also no measurement of what the participants ate, except for macronutrients and some micronutrients.  Calorie intake was identical and macronutrient ratios were fairly similar although the Paleo group had slightly lower protein, fat and SFA; and slightly higher carbohydrate and MUFA.

Energy (kcal)
Pro:Fat:Carb (%)
Sodium (mg)*
Potassium (mg)*
Calcium (mg)
Fibre (g)**
* The figures for sodium and potassium intake in the Paleo group are in mmol (the unit they reported), while those for the ADA group seem to be in mg
** The fibre intake in the Paleo group is what you would expect with 3000 calories, a high carbohydrate diet and most of the carbohydrates coming from fruit and vegetables.  The fibre intake in the ADA group is very low and if it’s correct, then the ADA group were probably eating a lot of refined grains rather than whole grains


Glucose control: fasting glucose significantly decreased in the Paleo group, with no change in the ADA group, although the difference between the groups was only P=0.3.  Consequently, fructosamine, a short term marker of glycemic control, significantly decreased in the Paleo group, with no change in the ADA group and a near significant difference between the groups (P=0.06).  HbA1c significantly decreased in both groups, despite no improvement in fasting glucose and fructosamine in the ADA group.  Both groups had minor improvements in insulin sensitivity (P = 0.1 and 0.09 respectively), although among the most insulin resistant participants, those in the Paleo group, but not the ADA group, had a significant improvement in insulin sensitivity (see figure below)

Blood lipids: both groups had decreases in HDL-C, probably due to lower fat intake.  The Paleo group, but not the ADA group, had significant decreases in total-C and LDL-C probably due to both the very low SFA intake and high fibre intake.  Triglycerides almost significantly decreased in the Paleo group, but not the ADA group, despite higher carbohydrate intake and no difference in weight loss and baseline triglycerides, which may be due to high fibre intake.

ADA Diet
Paleo Diet
P Value Between Groups
P Value
P Value
Weight (kg)
Systolic blood pressure
Diastolic blood pressure
Mean arterial pressure
Fasting glucose (mmol/l)
Fructosamine (mg/dl)
Insulin sensitivity
Total-C (mg/dl)
Trigs (mg/dl)
HDL-C (mg/dl)
LDL-C (mg/dl)
Creatinine clearance
Urine K/Na**
Urine pH
Urine Ca/Creatinine
* Bold = p < 0.05.  Underline = p < 0.10

** The urinary potassium:sodium ratio reflects dietary intake, and therefore can be used as a marker of compliance.  “Calculation of potassium to sodium ratio confirmed that all the patients, except for one, on the Paleo diet were compliant with the diet”

Lastly, after the trial was over the participants seemed to revert back to their old diets (based on urine sodium and potassium) and consequently the reductions in cholesterol and HbA1c were also reverting back to the usual level.