Self experimenting with a 3 day fast

I thought it might be interesting to monitor and post some data about a 3 day fast. Originally I wanted this to be a much longer post, a generic recipe to monitor fasting. But as I was writing it I realized it was boring, and sometimes I was talking out of my ass. So here are some relatable tables with biomarker data instead.

If anything this might hope more bloggers to document their self-experiments. I think the state of medical research is such an unreplicated and unexplored mess in many areas (this included), that n=1 studies done with some amount of care might be surprisingly useful, at least if enough of them are done.

I originally wanted this to be a 7-day fast, but due to feeling blue, I dropped it to 3-days (roughly 72 hours from the last to the first meal).

BY fasting I mean <30kcal a day (so basically ~1% of my normal caloric intake), allowing for stuff like a can of coke zero, xylitol gum, omega-3 supplements, vitamin A supplements, a pinch of cream in the 5 o'clock tea and so on.

Lab tests

I went to a lab 14 hours into the fast, to get my normal "fasting" values for the "before" measures. Then, 2 nights after the fast ended, I went (fasted for, again, roughly 14 hours) and did the "after" blood values. I waited 2 days since I wanted to be, to some extent "back to baseline". Ideally, I would have done this at further away intervals too, say 5, 10 and 20 days afterwards, to see how long the effects last and what sticks. I will probably do this at some point in the future. My guess personal guess is that they mostly "level off" in 1 to 8 weeks.

So what data did I get?

Test Before After
Uric Acid [mmol/L] 354 289
homocystine [umol/L] 13.1 12.7
Fibrinogen [g/L] 2.67 2.4
CRP [mg/L] 3.41 0.5
HDL [mmol/L] 1.73 1.75
LDL [mmol/L] 2.33 2.45
trigs [mmol/L] 0.6 0.8
ALSO [kIU/L] 27 27

Well, I also tested some hormones and calcium, but the lab I went to gave me some nonsensical values (i.e. two tests that ought to show the same thing showing different values) and I chose to exclude sensitive blood works where having a decent lab is important. Since I got an overwhelming amount of hints these guys were incapable. So I'm only including and trusting the thing that are hard to screw up.

Even so, there are literally a few dozen more things I ought to have tested but didn't, mainly because this was my first try at doing this after 1+ years. So I wanted to keep it budget and metal-load friendly.

I got the ldl/hdl/trig combo because they were offering that lipid panel for free. I don't think these values are super useful but if you are at the age and lifestyle where you do, my PSA is to look into monitoring more in-depth biomarkers (e.g. apoB)

ALSO is an anti-strep antibody. My pet theory was that maybe this fluctuates and is a good direct marker of adaptive immune activity. But either the lab is bad and instead of saying "quantity insufficient for assay to determine a value" it just says "27" every time ... or the level of these antibodies are much more constant than I thought. The latter explanation is weird, I'd expect that at least my background level of strep exposure to influence the value a bit, so I tend to believe that the lab messed up. Anyway, nothing to see here, and I probably won't bother with this in the future and instead will do more rigorous tests.

The rest of the tests are very encouraging and I think it's a very good showcase for why you should try fasting, now. They motivate me to scream at people (more) that they should drop whatever nutritional and lifestyle intervention they are doing and start a habit of occasional fasting first. Then again, I remind myself, n=1.

C-reactive protein and fibrinogen are proxies for "non-specific T cell activity, plus internal clotting, plus a bit of liver function, plus a few other things that are even less agreed upon". Many clinicians and studies rely on them as good markers of chronic inflammation, at the least they are good in that one can find good reference ranges and they aren't heavily confounded by too many factors.

I'm honestly not sure what the metabolic pathways involving homocysteine are, I know the "common" explanation is that it damages arterial walls, but I am also close to convinced it's probably just an unfounded rationalization of some weak correlations (which is just my starting point for any simple mechanistic claims regarding CVD). The link of "more==bad" seems fairly strong and the link to chronic inflammation is also fairly strong, so ultimately that's all I care about.

Uric acid I measured in part out of personal curiosity, since my diet is heavy in things that might risk increasing it... turns out it was quite high. Not outside of "reference ranges", but I'd sleep much better at night if it was half or at least 2/3rd of that value. It's also associated with various CV issues, stroke, gout, arthritis, chronic inflammation... the lot.

The levels of fibrinogen and homocysteine decreased by a bit, but they were already quite low.

The levels of uric acid decreased by, well, not quite enough to make me comfortable, but still significantly. I was actually curious how this will behave given all the cell death and lysosomal activity that may have lead to more. So at least here's one tentative finding that I couldn't find in any article:

I doubt this finding helps anyone out, but who knows, maybe someone with gout or arthritis wants to try fasting, is concerned about their UA levels, stumbles upon this and is swayed to give it a go? If you are that person, please do your own monitoring, since this might not generalize to you or the finding might be spurious.

Finally, C-reactive protein dropped, by a fuckton. This is really good, at least in that I can find no claims that CRP is anything but "the literal devil" in any medical literature. So, ahm, yey! ... ? I didn't expect I could get this value so low.

Fasting at first sounds like a weird proposal, I'm not sure these biomarkers will help to sway anyone. But at least they might convince some more people to monitor what happens when they fast. They might even convince the most curious and hypochondriac to try it.


Fasting will cause you to lose weight. Generally speaking, this is good, in my case it almost certainly is. What I'm looking for here is preserving as much bone mass as possible while increasing the muscle/fat mass ratio.

All measurements will be made using a cheap electrical impedance myography (EIM) scale (XMTZC05HM). About the same time during the day (~10AM), after waking up and doing some mild aerobic exercise then resting for a bit.

marker day 1 day 2 day 3 2 days after
muscle [kg] 67.29 66.89 66.17 66.7
fat [%] 20.2 19.7 18.9 19.6
water [%] 54.8 55.1 55.6 55.1
bone [kg] 3.62 3.6 3.56 3.59
(m+b)/f 3.51 3.57 3.69 3.58

I guess you could call these results promising. Check the last row (muscle plus bone mass over pct fat) for a rough endpoint that I'm trying to improve here. The "real" endpoint ought to be more complex and take into account water pct and total weight and maybe a few more things that this scale doesn't even pretend to measure.

But given how shitty the scale is, it doesn't really matter. What I take from these measures, combining them with my limited knowledge of body composition, metabolism and EIM, are the following ideas (significant == more than would be expected from daily fluctuations):

  1. I'm losing weight [99.9% confidence]
  2. I'm losing significant amounts of both fat, muscle mass [90% confidence]
  3. I'm losing a significant amount of bone mass [70% confidence]
  4. These proportions are decently healthy, they are equivalent~ish to what I'd expect on most diets. [50% confidence]
  5. But they are not ideal, and if I wanted to lose (small) amounts of weight I think a better route would be preferable [50% confidence]
  6. The decrease in fat mass remains once the fast is completed provided I eat the same amount as before [40% confidece]
  7. The fast improve muscle/fat ratio long term [60% confidence]

The one weird thing here is the change in bone mass. Either I'm losing 60 grams of bone during 3 days (even though I'm supplementing NaCl and Mg) or the scale is bollocks. But it's at least prompting me to look a bit more into fasting and skeletal changes (read: potential for regeneration from arthritis)

Glucose challenge test

A standard test for assessing insulin response and production is the GCT. Depending on the source this is done with a fixed amount (50 or 75g) of glucose in a 20/80 solution with water or with 0.5-1g/kg of body weight of glucose. Then serum glucose is measured at a few fixed intervals afterwards.

Both approaches are faulty since not all body mass is created with an equal insulin response; Nor is feeding a fixed amount of glucose sane, given that variation in how much people eat ranges from 1,000kcal to 12,000kcal a day (think the guy that played Tyrion vs the guy that played The Mountain).

But the test can be used as a relative comparison for your overall metabolic health. If the values are extremely weird, it can also indicate a problem even as a one-off.

Nowadays humans eat a lot of fructose and in the cases of some diets (e.g. mine) galactose. Also, drinking 50 or 75 or 100g of glucose is kind of unhealthy, and it tastes blergh.

So I'll instead drink a coffee with milk, eat a disgustingly nice fresh chocolate doughnut and some chocolate muse (exact same brand and portion size each time). I wrote down exactly how much sugar was in these, I remember it being equivalent to roughly 112g of sugars with a roughly 60/30/10 glucose (including that from starches) /fructose (including that in sucrose) /lactose spread. But I can't find my notes to save my life and this is going off 2 months old memories. I ate this after ~16 hours of fasting, 2 days before and 3 days after the fast.

I should have also measured at some "normal" intervals like 0,15,30,60,120 minutes... but I kinda messed up the first one, and had to try and replicate it the second time, so the measurement times are weird, sorry.

blood glucose [mg/dL] - before blood glucose [mg/dL] - after Time
98 89 before eating
94 100 minute 0
168 128 minute 27
116 98 minute 55
105 93 minute 62

Note to self: I should really just bite the bullet and get a continuous glucose monitor.

I'm seeing exactly what I expected, insulin response (inferred from the trend of glucose) goes from "you're not diabetic and doing decently for someone your age" to "so good I'm sure most doctors aren't aware you can get this kind of curve without exogenous insulin".

As far as I know, the lowest risk of mortality for my age bracket (and for people in general) is associated with somewhere just about 90 mg/dL of blood glucose.

So in theory the after-fasting baseline is ideal if a tiny-tiny bit below. But I think there's a bunch of problems with the study above and similar ones, so I don't put much faith in it. I didn't have BHB tests at this point, can I assume it was higher post-fast and maybe that's why the glucose was lower? Maybe, circulating trigs increasing also point in that direction. But without ketone measurements and a continuous glucose monitor, I can't tell much.

Speaking of which, is it ok for glucose to be this low after a meal?

I basically ate a desert and my peak value would have been considered healthy fasting glucose... that in itself just tells you how fucked up the medical system is, and why there is an epidemic of obesity, NAFLD, metabolic syndrome and diabetes. But still, it normalizes very quickly.

I tend to think this is ok, the body doesn't have much to do with glucose above ~100, and it can't work optimally for a long time with glucose < 85 or so (unless in ketosis). So in theory this seems like an "ideal" reaction to eating. But it's certainly not what you'd see in 99% of individuals. So buyer beware, I think this indicates an improvement, but depending on how much you trust "average is right" you might disagree.

Wrapping up

I want to try and write more posts that are "descriptive" of reality, or at least of conceptions about reality, rather than theory-building. I guess the theory is just a way to summarize descriptions, but building it kind of constraints the content and the topic.

I also want to write more about what I'm doing, rather than to try and write about a "higher level" based on my own experienced. Writing about a "higher level" that generalizes better might force me to be more rigorous, but it's also more tiring and feels dishonest in a way.

Will this post be interesting or useful to anyone? I don't know, if it is, let me know.

I think the takeaways here, IF (and this is a BIG IF) my data generalizes to you, are:

More broadly, maybe this is a push for you to experiment with fasting and at least see the benefits for yourself, or serves as inspiration for your own fasting monitoring protocol.

I might try posting a more in-depth analysis of a longer fast later, but that in part hedges on whether or not I'm able to set up the right conditions for maintaining and monitoring such a long fast. I'll almost certainly do a more in-depth analysis of a 3-day fast, to replicate these findings and add more biomarkers, as well as use a more reliable lab.

Published on: 2021-09-06



twitter logo
Share this article on twitter
 linkedin logo
Share this article on linkedin
Fb logo
Share this article on facebook