Dogma Destruction: Phinney’s Dodgey Ketogenic Cyclist Study Revisited

Cycle Ergometer with Metabolic Cart

Please note: This article contains the kind of language you won't hear on a Wiggles DVD. If you dislike strong language, please close this page immediately.

A reader by the name of Jake S recently wrote to challenge my critique of Stephen Phinney's infamous keto cyclist study. I've reprinted the subsequent exchange below. I present it, not merely for entertainment (although the cocksure and audacious nature of people who truly have no clue what they're talking about can indeed be a highly amusing spectacle), but because there are some valuable lessons to be learned  by scrutinizing Jake's irrational communications.

Jake S writes:

I read your criticism of Phinney's experiment on low carb cycling/exercise.

Your criticism boiled down to

1) Familiarisation accounts for increased performance (ignoring the fact that riding a bike is as easy as....well)

2) Some other guy with a PhD thinks, off the top of his head, the increases can't be right.

Powerful stuff Anthony.

Anthony replies:

So you deny the well-documented phenomenon of familiarization?

And you also deny that it applies to endurance activities, not just those requiring complex motor skills?

Powerful stuff Jake.

Try writing back when you have some intelligent commentary to offer...

Jake replies:

Documented where? Citations? And if documented, are the instances documented readily transferable to stationary bike riding?

I'm not saying it's not there just because you can't cite off the top of your head. That would be game playing on my part. Your two objections just seemed a bit flimsy for me....

Phinney's study indicates that there's a strong primary facie case that a "decent" level of physical endurance can be achieved with zero carbs. The fact that it (may) fall short of the peaks achievable by a high carb fuelled endurance is not really important, is it?

Attia documents impressive physical improvement on keto (more mental clarity and very good endurance). He claims he couldn't replicate this on the variety of carb high diets he tried: gained weight, felt sluggish etc. Does this not indicate a variety of optimum diets for humans?? Cardio endurance is only one factor right? His lipid profile is great and his visceral fat was measured as the lowest the guy measuring it had ever encountered (unless he's lying). Mike Arnstein was an aspirant runner who couldn't break sub 3 hours trying everything he could. Switched to all fruit diet and went 2:37 in his mid 30s. There seems to be an issue with mixing fat and carbs for some people - resolved by all fat or all carbs. From scanning the literature (very superficial look) both these diets reverse insulin resistance, so a noted mechanism is in place to account for their respective improvements.

Loved your response to Durianrider by the way. What a c*nt that guy is.

Best wishes

Anthony replies:

Harley "Durianrider" Johnstone is indeed a most repugnant individual, but your own email demeanour sure leaves a lot to be desired.

I've posted what I believe to be the most exhaustive online critique of Phinney's widely-cited cyclist study to date (if anyone else has posted an even more detailed critique, by all means, someone please send me the link), yet you snidely reduced my entire critique down to two intelligence-insulting bullet points that don't even begin to reflect the breadth of my commentary.

Not content with that, you then conclude with the insulting snipe "Powerful stuff Anthony."

I'm not sure just what kind of a person you are in real life, but when I receive an email like that from someone I've never had any prior communication with and therefore cannot have possibly provoked, it strongly suggests to me that the person in question is an obnoxious, belligerent, condescending dick.

When they then proceed with a proud display of their remarkable ignorance on the very topic they are challenging me, my impression of that person drops even further.

By all rights, I should ignore you and get on with other stuff, but I think an analysis of both your communication style and your woeful ignorance on clinical exercise research and cycling in general will serve as a useful case study for my readers.

So let's kick off with Phinney's study.

Your attack on my commentary on Phinney's study is a gross over-simplification - you read my detailed analysis of his farcical trial, clearly got irritated at my disputing of your cherished beliefs, then wrote to me in a most smart-assed manner, distilling my entire commentary down to two stupid little bullet points.

Let's start with Stupid Little Bullet Point #2:

2) “Some other guy with a PhD thinks, off the top of his head, the increases can't be right.”

With this comment, you've not only trumpeted your ignorance but managed to snidely deride, not just myself, but an accomplished exercise physiologist.

Way to go, Jake.

For the record, I never said anyone should dismiss Phinney's study because "Some other guy with a PhD" thinks it "can't be right". This "Some other" guy's name is actually Andrew Coggan, a prolific researcher who is a damn sight smarter than you and Phinney put together*. I quoted him because he's a bright guy who was one of the first to publicly call out these flaws, and there's nothing wrong with giving credit where credit is due. In fact, I think it's the decent thing to do.

1) “Familiarisation accounts for increased performance (ignoring the fact that riding a bike is as easy as....well)”

Freaking brilliant. You write to dismiss this possibility, but now you're backing off and admitting it's a possibility and asking me for citations. Which tells me you never researched the phenomenon of familiarization in the first instance. Instead, the possibility that familiarization affected the results just didn't "sound right" to you, so you figured you'd write and tell me I was wrong.

Yep, another joker who writes to tell me how wrong I am based on his "feelings", "hunches" and research he's never even viewed.

Here's a free tip: Research the bejesus out of something BEFORE writing to dispute something someone said.

And when you finally feel knowledgeable enough to comment, write to them in a civil manner instead of that an all-knowing twat who over-simplifies and makes uncalled-for comments like "Powerful stuff Anthony".

Otherwise, you come off like a certain banana-gorging troll...

Oh, and those citations you wanted?

Hmmm, I guess a lot of people still don't know about Pubmed ...

Here are some examples of clinical trials in which researchers examined performance across three subsequent endurance tests in the lab setting, and detected significantly greater performance improvements from test 1 to test 2, as compared to that seen from test 2 to test 3.

Schabort EJ, et al. A new reliable laboratory test of endurance performance for road cyclists. Medicine & Science in Sports & Exercise, Dec, 1998; 30(12): 1744-1750.

Hopkins WG, et al. Reliability of power in physical performance tests. Sports Medicine, 2001; 31 (3): 211-234.

Laursen PB, et al. Reproducibility of the cycling time to exhaustion at .VO2peak in highly trained cyclists. Canadian Journal of Applied Physiology, Aug, 2003; 28 (4): 605-615.

And while your initial email ridiculed the idea that test familiarization could ever occur in an 'easy' activity like cycling (you clearly have not done much cycling!) ... the above studies were all conducted using ergometers.

What does one do on an ergometer?

Stationary cycling.

Not doing too well there, Jake.

Incredible new discovery: People use stationary bikes for ... stationary cycling!

Your woeful ignorance of cycling is worth expounding on, because it provides a useful example of how people arrive at erroneous assumptions based on erroneous opinions and ideas they hold in their own heads.

Compared to activities like gymnastics or boxing, where your body moves every which way as the activity is performed, road cycling is a relatively one-dimensional activity. Your body remains positioned centrally over the bike, and the only major movement of note occurs at the hip, knee and ankle joints. And as the lower limbs move around these joints, they do so primarily in a vertical plane.

Which is where you make your first big error. You assume - based on what, I'm not sure - that test familiarization is limited to movements requiring sophisticated neuromuscular co-ordination. You either forget or are completely unaware that there are numerous other performance variables that could affect familiarization.

The one I'll focus on here is psychological, because it's extremely powerful but often overlooked.

If you knew anything about competitive cycling, you'd know that cyclists often do "reconnaissance" rides of the routes they will be riding in an upcoming competition. Someone like yourself, who thinks riding a bike is pretty much all about staying upright, turning pedals in a circle, and little else, probably wouldn't understand why, so let me explain it to you.

I ain't no pro cyclist, but I've clocked up quite a few bike miles in my time on Planet Earth. Yet no matter how fit I'm feeling or how well my training/eating/sleeping is going, the first time I tackle a new climb my time sucks compared to all my subsequent outings on that same climb.

In fact, when I do that same climb only a day or two later, my time obliterates the one I achieved on my previous outing.


Well, it obviously isn't any remarkable increase in my fitness, because such an increase simply could not occur in such a short time frame.

What can improve in that short time is my psychological approach to that route. It's all well and good to look at a route on a map, but it's not until you get out there on your bike that you start to truly understand which gear to use in what section, when to put the hammer down, and when to ease off a little. The first time you ride a route, you might be fooled into going too hard too soon, or leaving it too late to really open the tap. And if other people who've previously done this climb have wanked on and on about how painfully difficult it is, there may be an intimidation factor that, on your first outing, makes it seem harder than what it really is.

But after riding the route again you now know where the steep bits are, and what the last few corners look like. In other words, once you are familiar with a route you can pace yourself much more effectively. Also, once you've conquered a daunting climb and "understand" it, any intimidation factor quickly disappears.

The end result of all this is that, even though you may not be physically fitter, and even though the climb itself has not physically changed in any way, your subsequent outings on that route become psychologically easier. And when your mind is fully on board with an activity, a little something called your rating of perceived exertion decreases, which often paves the way for bonafide physical performance improvements (and no, I'm not going to launch into a separate discussion on RPE and supply you with references ... it's real, it's well researched, and the information is readily available on the interwebz for anyone non-lazy enough to look for it...).

Cadel Evans made sure he got familiar with Adelaide's Corkscrew Road before the 2014 Tour Down Under; the result was this truly stunning performance.

Now...Phinney's study used what he described as competitive amateur road cyclists, but little detail was supplied about their usual type of training or competitive activity.

What we do know is that the test employed in the study was a cycling-to-exhaustion protocol performed on a stationary bike.

While treadmill running and ergometer cycling to exhaustion are very common tests in the clinical exercise physiology setting, they actually bear little resemblance to real-life running and cycling events.

There is NO officially-sanctioned cycling event in which the participants are told at the starting line, "OK guys, when the flag drops, ride at 65% of your MaxVO2, and ride at that pace for as long as you can. Last man standing wins!"

In real life events, there is a clearly defined and pre-determined course. If it is a circuit-type course, the usual scenario is that course must be completed for a certain number of laps, and the first person across the line after completing the requisite number of laps wins. If the course is a non-circuit route, such as that seen in most classic and stage races, the goal is equally simple - get to the finish line first.

If it is an individual event, such as a time trial, the goal is to get from the starting gate to the finish line in the quickest possible time. No deceptive tactics, no breakaways, no bullsh*t - just get out their on your own, haul ass and try and clock the day's fastest time.

Now... when you take a bunch of cyclists and ask them to perform a test that most if not all of them have never done before, and which bears little resemblance to their usual style of riding and competing ... do you think it's possible they may undergo familiarization as they progress from the initial test to the subsequent test?


You have a bunch of guys whose normal cycling activity occurs outdoors, involves changing routes and scenery, involves the selection and use of different gear ratios during a ride, and also involves varying tempos and levels of exertion. But now you are sticking them indoors on a piece of equipment that goes nowhere no matter how fast you pedal. You are telling them to sit on the spot and pedal that piece of equipment at a fixed pace, in a single gear, for as long as they possibly can.

In other words, you are now asking them to perform an unfamiliar activity.

And in Phinney's study, they didn't just have a week to mentally process their novel new testing procedure, as in most familiarization studies - they had an entire month.

In contrast to your claim my criticism of Phinney's study is comprised of "two objections", I actually discussed a whole host of problems that render Phinney's study essentially worthless.

--No familiarization test. While you think this is unnecessary, it is actually crucial and a standard procedure for any properly conducted clinical trial utilizing physical performance tests.

--No random assignment to treatment. It's also standard procedure when comparing 2 different diets in crossover fashion to randomly assign the participants to follow the diets in different sequence. In Phinney's study, this means that some should have followed the mixed diet first and then the ketogenic diet, while others should have started out on the ketogenic diet then followed the mixed diet.

Why is this important? Because if a difference in a performance variable is noted among the two diets, we can be confident it was a result of the diet itself and not an artifact of the switch from one diet to another (it's no big secret that switching between starkly contrasting dietary regimens can result in temporary perturbations in biochemical/physiological parameters).

--Uneven periods on each diet: The cyclists were observed for only one week while following a mixed diet, but four weeks following a ketogenic diet. What would the results have been if they followed both diets for an identical period? Thanks to the poor design of Phinney's study, we'll never know.

--A decline in the cyclist's sprinting performance. This was not even mentioned in Phinney's original papers; he only mentioned it in a throwaway comment some two decades later in a 2004 discussion paper.

--Patently misleading claims about the ketogenic diet's effect on endurance performance. Phinney's study is cited ad nauseum - by both Phinney and low-carb advocates - as 'proof' that ketogenic low-carb diets do not hamper endurance performance.

Making such a claim from this study is blatant bullsh*t.

There were a grand total of five cyclists involved in this tiny study. While two did indeed pedal for a significantly longer duration on the subsequent time-to-exhaustion test (one experienced a non-significant increase), the remaining two experienced hefty declines in their performance (-48 and -51 minutes).

And remember, that was despite the very real and likely advantage of familiarization from the first test to the second.

Once again: Taking a study in which 40% of the participants experienced significant performance declines after following a ketogenic diet, and then claiming this same study showed ketogenic diets do not impair endurance performance is complete bollocks.

So how does Phinney get away with this? Easy. Forget the individual results, and instead add the combined times together and calculate an average time-to-exhaustion figure. Bingo! The resultant figure makes it look as if no decline in performance was observed!

Yep, while in a league of their own when it comes to statistical chicanery, epidemiologists are hardly the only researchers who can make numbers say whatever they want...

Phinney's study was small, poorly conducted, and poorly reported. And even if he pleads ignorance based on the fact that the study was conducted some three decades ago, and clinical trial conduct wasn't as advanced back then as it is nowadays, the fact remains he keeps pimping this study as proof low-carb diets are the schnizzle for exercise despite a host of conflicting research from scientists who, unlike Phinney, have no financial interest in the low-carb paradigm.

Phinney has also had some three decades to conduct a trial in which the numerous flaws of his earlier effort have been corrected - but he has done no such thing.

Despite this, people like you Jake get angry at me - not Phinney - for highlighting his untenable claims. That's because you're hopeless dogmatists who value preconceived dogma above the actual facts.

If Phinney's study made the exact same claims about a vegan diet, and I dissected his claims in the exact same manner, you low-carb jokers would all nod your heads in smiling agreement. I know this for a fact because years ago when I used to confine the bulk of my research analysis to dissecting fallacious claims about cholesterol and dietary fat, low-carbers would regularly write to me telling me how super-freaking-awesome-and-lovable I was. That completely changed the day I posted an article back in 2005 stating low-carb diets offered no 'metabolic advantage' - that the overriding non-genetic determinant of how much fat you gain or lose is how many calories you take in versus how many calories you burn off.

The subsequent outpouring of unbridled hate following that post was quite disconcerting. Never mind that I was actually following a low-carb diet at the time, and that the post in question was actually defending low-carb diets, and that I therefore had no financial or ideological motive to critique the metabolic advantage theory.

Nope, never mind any of that. I had just disputed a key tenet of the Church of Low-Carb, and for that I was now persona non grata.

That experience and all that followed, along with my own personal research into evolutionary psychology, taught me that most people - despite their insistent bleatings to the contrary - really don't give a flying act of fornication about the truth. What most people really want is that warm, fuzzy, reassuring and self-validating feeling that comes from reading something confirming what they've already become comfortable believing.

Well, here's a new flash for you Jake, and the rest of your dogmatic low-carb cohorts: When I was born, no-one ever asked me to sign a statement saying I would endeavour in life to always tell people what they want to hear. And if such a contract was presented to me, I probably would have dribbled on it and told its bearer in infant-talk to go screw themselves.

My purpose when I read the research is to find out the facts. And when I share my findings with others, my purpose is to relay these facts truthfully, not to piss in their ear.

If that is disagreeable to you, then here's another free tip: Don't visit my site. Because if you only like reading stuff that supports your preconceived beliefs, then my site is really going to piss you off.

Before I close this off and add your email to my blocked list, I can't let your comments about Attia and Arnstein go without comment.

You are more than happy to uncritically accept the unverifiable, anecdotal and experiment-of-one claims of these two gentlemen, yet demand I provide you with research citations proving the existence of test familiarization - a phenomenon that anyone who knows anything about clinical research should be fully aware of.

Powerful stuff Jake … LOL.

*For those interested, Andrew Coggan PhD has co-authored a highly-regarded book on using power meters titled Training and Racing with a Power Meter.


Anthony Colpo is an independent researcher, physical conditioning specialist, and author of the groundbreaking books The Fat Loss Bible, The Great Cholesterol Con and Whole Grains, Empty Promises.