By Robert Verkerk PhD, scientific and executive director, ANH-Intl

You’ll be familiar with the old adage – one frequently applied to things nutritional: “everything in moderation.”

Does that also apply to heavy metals in your fish, pesticides on your blueberries, or transfats in your gluten-free flapjacks? Of course not! There are some things that just aren’t good for us. And there’s currently a big debate as to just how much and what kinds of carbohydrates we should be eating for optimal long-term health. Unlike fats and proteins that we can't live without, carbs are - technically speaking - optional. Yet most humans on the planet rely on them as the primary source of energy. And if we choose to restrict our carbs, what should we replace them with, if anything?

The latest scientific call that seems intended to reinforce public health messaging about carbohydrate intake is directed fairly and squarely at the rapidly growing low carb community. It’s an open access paper published last week in Lancet Public Health that suggests that low carb diets – as well as high carb diets - contribute to early death. The take-home? Eat carbs in moderation, so that they make up 50-55% of the energy of your diet, just the way governments are currently suggesting.

This advice is ironic when we know the majority of the population in Western countries are already doing this yet they’re suffering early, preventable deaths at the hands of obesity, diabetes, heart disease and cancer.

The low carb community, of which we are a part, is comprised of large numbers of consumers, doctors, scientists and even celebrities who have experienced or witnessed spectacular health turnarounds after reducing their carb intake. That includes remission from diseases like type 2 diabetes that was, until recently, widely labelled as 'incurable'. Supporting this experience and observation is a growing armoury of scientific evidence from diverse disciplines ranging from biochemistry to evolutionary biology that goes a long way towards explaining some of the key mechanisms at work that help people to regain their health.

The truth is confusion in this area is rampant. That’s not least because there are, as yet, no clear, universally accepted definitions for what is meant by a ‘low carb’ diet or a ‘low carb high fat’ (or LCHF) diet. That's despite Richard Feinman PhD from State University of New York Downstate Medical Center in Brooklyn, New York, and 25 other scientists and clinicians attempting to clarify definitions in their seminal 2015 publication.

The Lancet Public Health attack on LCHF

Understandably there has been a reaction by some of the leaders in the LCHF community, with the likes of Richard Feinman leading the charge.

While the ultimate aim is to have the paper retracted, this we think is a tall order given the scale of interests involved in maintaining the status quo. We shouldn’t forget that most type 2 diabetics who adopt low carb or 'keto' diets are able to come off medications, such as metformin, Glyburide and Januvia. Type 1 diabetics generally substantially reduce their insulin. And it’s a Lancet spin-off journal we’re talking about.

Are the findings of the new Lancet Public Health study reliable?

In short, like Dr Feinman and others, we say no. And that’s not just because we don’t like the findings or the paper’s conclusions. Predictably, the study's conclusions have been reverberated around the world. They’ve made headlines that include, Low carb diets could shorten life, study suggests (BBC), Both low- and high-carb diets can raise risk of early death, study finds (Guardian), Low-carb diet linked to early death, medical study suggests (USA Today), amongst many others.

What kind of study is it? In essence, the study has two parts. The first is a survival analysis which measures all-cause mortality in the ongoing prospective observational Atherosclerosis Risk in Communities (ARIC) cohort study of 4 different US communities and then attempts to relate this to self-reported intake of carbohydrates. The second part is a meta-analysis which compares the ARIC data with 7 other cohort studies, including the multinational PURE study published last year that included findings from over 135,000 people on 5 continents. After 7 years of follow-up, the authors of the PURE study concluded that high carb diets led to the highest mortality rates and those consuming more fat (albeit only 35% of daily energy) were 23% less likely to die than those who consumed least fat (10% of daily energy). It couldn’t exclude the possibility that less carbs and more fat were better still, following a linear relationship, so it was generally good fodder for those wanting scientific support for low carb diets.

The PURE study therefore put some high profile noses out of joint. Not least were the scientists that have been primarily responsible for informing current dietary guidelines, including eminent epidemiologists like Professor Walter Willett from the TH Chan School of Public Health at Harvard.

In many ways – the latest Lancet Public Health study can be seen as the establishment response to PURE – or what some might argue was the incorrect interpretation of PURE. It’s also no surprise to find Walter Willett, the architect of the original questionnaire used in the ARIC cohort, listed as a co-author.

Our concerns

We are independently preparing a response to the senior author of the paper to which correspondence is directed, Dr Scott Solomon from the Cardiovascular Division, Department of Medicine, Brigham and Women’s Hospital. So we won’t get bogged down here with the detail of our scientific complaint. However, we’ve broken down our concerns into 5 main areas, and below we summarise in lay terms some of our main concerns:

  • General methodological problems. Unmatched groups with different carbohydrate intakes, lack of adequate definitions of low carbohydrate diets (<40% of energy intake is insufficient), measurement errors from food frequency questionnaires, insufficient consideration of confounders
  • Potential unrecorded changes in dietary composition during follow-up period. The current study on the ARIC cohort included just two snapshots of self-reported carbohydrate intake based on Willett’s 61-question food frequency questionnaire (FFQ). These snapshots were taken in Visit 1 (1987–89) and Visit 3 (1993–95), yet there were four other visits in which other data were collected, most recently in 2016-17 (hence the cited 25-year follow-up period). There are many questions about the accuracy of the FFQ, there are inadequate reliable data on the quality of foods that have been substituted or how the composition of commercially available foods has changed over the last 25 years
  • Confounding. The authors have attempted to adjust for age, sex, education, waist-to-hip ratio, smoking, physical activity, diabetes, location and total energy intake. But there are numerous other confounders, including glycaemic index/load, phytonutrient, fibre and fatty acid profile, and the degree of processing in foods consumed that have not been factored in. Given what is known about the influence of these unconsidered factors, they alone could have exerted a major influence on the results, and been more important than total carbohydrate intake that the authors suggest is the prime driver of the associations found. These unconsidered factors could influence The authors consider the (positive) influence of plant-based over animal-based diets so they know that influences other than carbohydrates might be highly significant, yet they apportion all of the primary effects to carbohydrate consumption.
  • Confusing association with causation. The big message from the authors is the U-shaped relationship between carbohydrate consumption and mortality. They suggest there’s a ‘sweet spot’ for carbohydrate intake and this sits somewhere in the middle, at 50-55% of total energy from carbohydrates. Eat more or less carbs than this and you’ll die more quickly, suggest the authors. Interestingly, the figures for increased death rates among high carb consumers come mainly from the PURE study which includes less developed countries, and the figures for the left tail of the U-shape come from the new analysis of the ARIC cohort, in turn from a Western (US) population. There are so many differences in the genetics, diets and lifestyles between these two groups and the adjustments carried out only a go part of the way in addressing them. Moreover, it's entirely wrong in our view to suggest, as the authors have done, that it is the total amount of carbohydrate that is responsible for the higher death rates among higher and lower consumers of carbs. It's nothing more than a statistical association born out of the limited dataset that may be based on inaccurately recorded information (especially in the case of the ARIC-influenced left tail)
  • Relevance of findings to the real world. The low carb consumer group in the ARIC cohort appear very different from modern, informed, health-aware LCHF consumers. The authors characterise them as being “more likely to be young, male, a self reported race other than black, college graduates, have high body-mass index, exercise less during leisure time, have high household income, smoke cigarettes, and have diabetes." Their higher educational status may be the only characteristic shared with the majority of contemporary LCHF, Banting or 'keto' aficionados. The substitutions made by a modern-day LCHF consumer are also likely to be a whole lot healthier than those in the ARIC cohort.

Take home

When the survival analysis plots for the ARIC data computed a U-shaped curve, you can only imagine the excitement among some of the authors, particularly those who’ve long supported relatively high carb, wholegrain, low fat, plant-based diets. They'd blown away the possibility that low carb is better than moderate carbs - and that there's a linear and positive correlation between carb intake and all-cause mortality. But that's only if the data were reliable and relevant.

The problem is that the ARIC data aren't representative. It’s a fudge. Two self-reported snapshots, nothing like enough data on other factors that could have influenced mortality – and all of it entirely unrepresentative of what modern-day, low carb consumers are eating or doing.

The problem with science in a world in which so much science is controlled by so few (i.e., big corporations, governments, the Bill and Melinda Gates Foundation, etc.), is that it’s very vulnerable to manipulation. Nutritional studies are notoriously difficult; unlike drugs, nutritional studies can’t compare the effects of eating against not eating. When we cut down one type of food, what do we substitute it with? And when we look at studies with very long follow-ups like the ARIC study, do we cater for all the changes that have occurred in our food production methods over the last two and a half decades?

Public health policy relies on big studies with big numbers, and increasingly, ‘mash-ups’ (meta-analyses) of multiple studies with data that can’t or shouldn’t be compared, like apples and oranges. These studies are often quite indifferent about the variable quality of data going into them. In the field of nutrition, randomised controlled trials (RCTs) and prospective cohort studies (PCSs) have become the almost exclusive sources of information for public health advice. Clinical findings and experience are dismissed as anecdote.

We might need to wait a bit longer to understand more about the health of those who choose to adopt low carb diets that include a diversity of plant-based nutrients and/or healthy animal-based foods, as proposed in our Food4Health guidelines. But there are some over-arching principles we consider to be entirely compatible with low carb, healthy diets:

  • Restricting or eliminating all highly refined (high glycaemic load) carbs and added sugars
  • Reducing intake of carb-rich (starchy) root veg (most notably potatoes) is a good idea for most people, unless they are extremely physically active
  • Focus on getting carbs, fibre and phytonutrients from above-ground veg (and limited amounts of fruits) to support the microbiome and a range of metabolic and disease-protective processes
  • Eliminating all processed meats
  • Where meats are included in the diet, ensure they are properly sourced and cooked without open-flame and/or high-temperature cooking (e.g. grilling/barbecuing, broiling, or roasting)
  • Restricting your carbs to less than 30% of total energy
  • Reducing overall energy intake compared with national averages
  • Increasing the duration between feeding intervals (intermittent fasting) compared with national averages i.e. for most adults, cutting out snacks and generally eating no more than two meals a day
  • Helping your body to become ‘keto-adapted’ and regularly move into a state of nutritional ketosis through a combination of both low carb, energy restricted diet (compared with the norm) and altered physical activity and lifestyle patterns.
  • Getting back to the messaging linked to the Lancet Public Health study: how can we blindly follow government guidelines when we know - without any need for survival analysis or statistical trickery - that adherence to them is closely associated with the chronic disease spiral?

We’ll keep you posted on our progress with our correspondence with the paper's authors.


Food4Health campaign

Volek & Phinney (2011) The Art and Science of Low Carbohydrate Living

Volek & Phinney (2012) The Art and Science of Low Carbohydrate Performance


Back to ANH-Intl homepage