Fitness Fallacies: Thought Errors in the Industry

There are various movements in nutrition, and one which is quite subtle and doesn’t really have a name, but can be as overly reductionist and obstructionist as any other, is the “Nothing Matters Diet”. Except calories, of course.

It takes an established, mostly accepted (beyond a few in the low-carb space) physiological fact (i.e., thermodynamics and energy balance) and turns it into a monolithic explanation and solution to any question of nutrition, diet, and health.

It is often predicated upon the most mind-numbing of Fitness Bro Logic: “Well, in case you haven’t noticed, Mr. Scientist, there is an epidemic of obesity and overconsumption of processed foods, ergo the solution is a diet of personal responsibility, nutrient-dense meat, and a calorie [fucking] deficit.”

The additional expletive provides the middle letter in the now infamous fitness industry abbreviation, ‘CFD’, the populist lowest common denominator of lobotomised opinion pertaining to the major public health nutrition questions of the day. It is just that simple! No wonder it is such an effective prescription at the population level.

It’s a paradigm that seems determined to view everything through the lens of ‘adherence’, as if the entire population is enrolled in one big giant weight loss intervention. Any considerations of nutrition, health, or wider social, economic, and environmental factors become relegated to mere sideshow events, often viewed through the lens of concepts like ‘personal responsibility’ and/or ‘willpower’. It deliberately seeks to downplay any – literally any – evidence associating diet, foods, or specific nutrients, with adverse health outcomes.

I realise I am taking a broad aim at the fitness industry, so I must begin with the #notallPTs caveat. And this is not a token caveat, there are plenty of people in the fitness industry who are strong voices for science, embrace nuance and complexity, and have as much disdain and frustration for some of the narratives, CFD merely one of several, that are ubiquitous in the fitness space. To distinguish, I’ll refer to the more moronic side of the industry as the ‘BroSpace’.

However, while I know that the issues I’ll elaborate on are not universal, they are endemic. And the reason widespread Fitness Fallacies are an issue isn’t because the fitness industry is responsible for dissemination of nutri-nonsense (although it has its fair share of that). In fact, it is the opposite: it’s because the vogue now in the fitness industry is to be “evidence-based”, and that term seemingly appears in the bio of every marauding internet warrior.

And while broadly this move to more evidence-based emphasis is to be welcomed, the issue here is that, because there is no barrier to entry, many people in the BroSpace seem to gravitate to any mix of snake oil salesmen, charlatans, or the more savvy ‘gurus’ who give their arguments a veneer of ‘science’. Which in the BroSpace is basically saying “studies show…”. What you end up with is a mini army of Stan Efferding’s, that obnoxious combination of critical incompetence and subject matter ignorance combined with total conviction of correctness. Never a good look.

On the back of a number of recent conversations about various nutrition questions with representatives of the BroSpace- saturated fat, sodium, vegetable oils and polyunsaturated fats, red meat, health improvements independent of weight loss – I thought I’d expand in this essay on some of the most common fallacies and reductionist heuristics that seem to pervade the fitness industry.

The Unicorn Avatar

I’ll start with this because in certain circumstances, this can be useful heuristic, but mostly it is deployed in inappropriate circumstances that largely relate to conflating individual vs. population risk, and over-extrapolation, both of which I’ll cover separately.

The Unicorn Avatar: health conscious, physically fit and active, and all of the dietary basics covered. They eat all their veg, get their fibre in, eat their oily fish, blah blah. They may also track macros and do all that CFD stuff.

Now, you might be thinking, “this isn’t really a Unicorn because these people exist”? Sure, they do: but they are absolutely not representative of the general population, and specifically the at-risk sections of the population.

This can be a useful heuristic when it is used in the context of an individual assessing their overall individual risk. For example, I’ve had two conversations recently with people whose health markers are all great, diet is top notch, sleep, physically fit, etc., but with one slightly elevated marker, specifically LDL-C and blood pressure, respectively. And the question was, in the context of these individuals’ diet, physical activity, otherwise spot-on health markers, would they be worried? And the answer from both was ‘no’. And that is totally fine for them to come to that conclusion: that is individual level risk assessment. It doesn’t mean the slightly elevated marker isn’t something to monitor, or that it will not become a risk factor worth treating in the future. It just means that, for now, keep on keeping on and focusing on the other positive aspects of health-promoting behaviour.

Where the Unicorn Avatar becomes a pothole is when you’re discussion risk factors, or associations with diet and risk – for example, red meat – and you get: “Well, what if someone is eating 10 servings of fruits and veg a day, 30g fibre, has PERFECT health markers and lifts weights and gets 9hrs sleep a night, do THEY need to worry about [insert exposure]? Exactly!

Whenever this Avatar becomes even remotely representative of the population, let me know. It’s an asinine, obstructionist strawman argument, which is used in the BroSpace to basically dismiss any diet-disease association, any association between a given food exposures and risk, and even the effect of specific risk factors, like LDL-cholesterol and hypertension.

It’s taken to various degrees of lobotomy in different circles, but perhaps the most nefarious is among low-carb/keto enthusiasts, many of whom waffle on about how the CRP is this or their triglycerides are that, and how nutrient dense their diet is, and brag about having an LDL-C of >300mg/dL. “Hey, look how high MY causal risk factor is, but all these surrogate biomarkers are fine!

But on the less obvious level of total clown behaviour, the Unicorn Avatar is used to pretend that associations between any number of exposures – red meat, sodium, saturated fat – don’t apply. They do. They apply at the whole population level. It just may not apply to a given individual. Which brings me to Bonnet Bee No.2.

Misunderstanding Individual Risk vs. Population Risk

Although Instagram may lead you to think that the perma-deficit ~10% body fat AbsBroChill person is the norm, nothing could be further from the truth. Moreover, this Avatar provides a fairly useless heuristic for thinking about the application of research, both diet and risk factors, on the whole population. This Avatar isn’t an answer to the questions posed by research findings, it’s an obfuscation of the realities of shifting the bell-curve of risk at the whole-population level.

The issue is that within the BroSpace literally everything – every exposure, risk factor, or relationship – is viewed through the lens of individual risk. And this is fine, as long as it is clear that the consideration is staying there. But that isn’t what happens: the individual risk heuristic is grossly over-extrapolated to assume universal applicability. And this is invalid thinking.

The reason is a principle articulated by Geoffrey Rose 35yrs ago: shifting a whole population into a lower risk category results in more individuals benefiting than shifting only high-risk individuals into a lower risk category. Here is a pretty image to illustrate this principle:


Individual risk is a matter for exactly that: the individual. Thus, the considerations of whether someone with good diet, lifestyle factors, and overall health markers, but with a single risk factor slightly elevated, or with high intake of a given nutritional exposure, needs to modify that exposure says absolute nothing about the effect of modifying that exposure across the whole population.

There is a simple point that underpins this fact: risk is not homogenous across populations. Even where the same nominal cause-effect relationship, i.e., hypertension and stroke or red meat and cancer, is the same, the risk between the exposure and outcome is relative to a number of characteristics in the population studied: background diet, wider lifestyle factors, environmental exposures, etc.

And all of these factors that get thrown around as the reason for no/low individual risk – vegetable and fruit intake, fibre, balance of fats, sleep, physical activity, adiposity, etc., etc. – these can all be controlled or adjusted for in statistical analysis. So if the nominal exposure-outcome relationship survives these adjustments, then there is no way to argue that risk is non-existent once Unicorn Avatar is applied to the consideration.

So truth is, I don’t care if a little salt is something A Bro needs to worry about. Sprinkle away, Bro. But I do care that sodium reductions of 15% in the UK population after reformulation was introduced in 2003 have contributed to a population-wide fall in blood pressure of 3.0/1.4mmHg, and a 42% reduction in stroke mortality (amongst other population-risk shifts).

Or let’s take another example: the NutriRecs red meat meta-analysis from last November that the Bro’s set up an OnlyFans account for. This analysis found that the absolute difference in colorectal cancer risk per 1,000 persons for reducing meat intake by two-thirds was 1 less diagnosis. That is individual level risk assessment: in the UK population that would scale to around 67,000 less colorectal cancer diagnoses. What makes you so sure you aren’t going to be the 1 less diagnosis? For total cancer incidence, it was 18 fewer diagnoses. Do the math.

More individuals benefit from shifting the whole population into a lower risk category than shifting only high-risk individuals. The purported invalidation of every diet-disease association, every risk factor, as irrelevant using an individual level risk heuristic is an issue the BroSpace is particular liable for. So, cut it out. Be really explicit when talking about individual risk vs. population risk, and if you see anyone using individual risk heuristics to validate an argument about the relevance of an exposure in the whole population, then you know they don’t understand the difference. Which brings us to Bonnet Bee No.3

Inappropriate Extrapolation and Over-reach

Inappropriate extrapolation is a common offence in the BroSpace, and often directly relates to what we’ve outlined above as conflating individual risk heuristics with whole-population risk reduction. But where it gets really egregious is when randomised controlled trials (RCTs) come into the picture. In this respect, if there is a cardinal sin that the BroSpace are particularly guilty of, it’s what is known as ‘simple extrapolation’.

Simple extrapolation assumes that the results of an RCT apply invariantly, i.e., holds true universally, in all other circumstances. And in the BroSpace, no other research designs exist other than an RCT, and the high watermark of critical analysis is seeing where a study sits on the evidential pyramid. When the results of an RCT come out, particularly if it fits the “Nothing Matters Diet” narrative, they are literally thrown around the BroSpace as conclusive evidence that whatever finding was observed therefore invariantly applies in all circumstances, to all people.

This assumes: a) that the effect in an RCT is automatically true (false assumption); b) that randomisation controlled for all variables (false assumption), and; c) that the calculated effect size is the true effect of an exposure (false assumption). That’s a lot of false assumptions to go making strong convictions out of, don’t you think?

The over-reach that comes from this kind of simple extrapolation underpins much of the “Nothing Matters Diet” rhetoric, but it amounts to nothing more than shoddy thinking.

Bertrand Russell wrote about a chicken. The chicken is fed every morning by the farmer, and thinks, this is great: I am fed every morning by this person. This is therefore true and holds invariantly. Until Christmas Day morning, when the farmer breaks the chickens neck. The chicken is dinner. The chicken didn’t understand that there was a wider socio-cultural context at play. Had the chicken known, this would have changed its assumption that a morning feed holds true in all circumstances.

Think like Bertrand Russell. Don’t think like a Bro chicken.

Imperceptible Reductionism

Bee Bonnet No.4 is also an extension of over-reach and overly simplistic extrapolation, but tends to manifest in a way that is methodolatry at its worst. It’s the affliction within the BroSpace that every conflict between observational research and an intervention study has to be reconciled in favour of the RCT.

Unless, of course, that observational research confirms the Bro-bias. In fact, it is rather hilarious to see some of the prominent voices in the BroSpace rip on nutritional epidemiology (displaying nothing but utter ignorance for the field in the process mind you), only to then get the caps on for “HIGHER SATURATED FAT REDUCES RISK OF STROKE!” Oh yeah, what was that ‘higher’ level, Bro? Less than 20g a day? Thought so…

But mostly it is the imperceptibly defective thinking which assumes that any result from a short term intervention informs long-term longitudinal effects and associations. “Ah, that 30yr cohort study found an association between saturated fat and heart disease? Well HERE is a 6-week RCT in otherwise healthy males consuming cheese which found no increase in cholesterol!” Just no.

This is everywhere. Studies on red meat in an otherwise 6% saturated fat diet, sugar in a hypocaloric diet [the fancy term for the aforementioned calorie deficit], studies on single-day acute sodium loading, studies on specific food sources of saturated fat, and studies in otherwise healthy participants. All combined into one big pile of “Nothing Matters” and proffered as some sort of model falsification for long-term diet-risk factor-disease relationships.

Of course, perhaps you’ve heard the famous HRT example? You know, the one where a prospective study found that HRT reduced cardiovascular risk in women, but then an RCT found an increase in risk? And obviously the RCT was right, right? Jog on: both studies were “right”, once they were analysed adjusting for the timing of initiation of HRT relative to the onset of menopause.

The problem in the BroSpace is that RCTs and meta-analysis are treated as invariant truths. And in nutrition science, nothing could be further from the truth.

Overly Simplistic Solutions

The effect of wider social, economic, and environmental factors which influence diet in huge sections of the whole population is now unequivocal. Yet, there is a flat out refusal within the BroSpace to engage with this reality, instead defaulting to anecdote-driven “personable responsibility” narratives. But this research, and the fact of its effect, exists whether anyone wants to believe it or not.

Uniformly, said proponents will have an IG profile pic of traps, abs, or even – should we be so lucky – an entire torso of manly muscle. Mention socio-economic factors, you’ll get told that anyone can learn to meal prep. Mention the fact that families in the lowest quintile of social deprivation in the UK would have to spend 75% of their income to comply with ‘5-a-Day’, and you’ll get told that they need to be more “fiscally responsible”. That it’s about planning, dedication, and goal-orientated behaviour. That we’re all one Jordan Peterson YouTube video away from self-actualisation.

To remind you of how much of complex web of factors influence the population, here is the Foresight Report map:

Having regard to that web of factors, you’ll excuse me if I’m pejorative about ‘calorie deficit’. Particularly where all of this complexity is dismissed out of hand by some plank who eats 6 meals a day out of Tupperware containers and lives in a gym.

The False Idol of Evidence

Sure, “evidence-based” is in vogue. I get it. But it means nothing. And under this guise, it is easy to get drawn into conversations that appear on the surface to be “evidence-based”, but in reality is just an exercise in using a few PubMed references to justify a position they want to hold anyway. Obfuscation, conflation of individual with population risk, overuse of the Unicorn Avatar, gross over-extrapolation, methodological prejudice, and complete dismissal of the wider factors beyond CFD that influence diet-health-disease relationships at the whole population, are all endemic in the fitness industry. And there is nothing “evidence-based” about any of it.