Going vegan: Preliminary metabolic analysis

About six weeks ago my spouse and I decided to go vegan. Our motivation was ethical. We worry about the suffering of animals under factory farming and the morality of killing, eating, or otherwise exploiting other sentient creatures. At the time some of my friends suggested that there would be health benefits as well. I was skeptical, since we’re both very active, lean, and already ate a clean diet that we controlled through rigorous calorie counting.

To my surprise, there was at least one apparent change: We both noticed a sudden drop in body weight over the first two weeks, despite no change in activity level or total calories consumed. This also seemed to be supported by data from our Fitbits. Just eyeballing the data, it looked to us like we needed to start systematically eating about 200 kcal/day over what Fitbit said was needed to maintain our weights. (We did, and our weight loss flattened.) This means that despite still expending roughly the same amount of energy (calories) per day, we needed to consume 200 kcal/day more in food to meet that energy demand. That is pretty shocking, since it’s around a 7-12% increase in consumption requirements just from cutting out animal products.

So, I dug into the data a bit deeper. Surprisingly, this analysis suggests (at least in my case) a bigger difference: around 300 kcal/day more needed, not 200 kcal. Although my informal analysis lacks the rigor needed to be fully trusted, it does strongly suggest that, relative to energy demands, a significant amount more needed to be consumed to maintain body weight.

First, a bit about myself. I’ve fastidiously counted calories consumed for the past five years and did so off-and-on before that. I eat almost exclusively food I make from scratch, and eat a relatively stable rotation of simple foods. I stick to a lot of oatmeal, rice, noodles, whole wheat bagels, fresh vegetables, and (before going vegan) boneless-skinless chicken breasts, eggs, and cottage cheese. I’ve also counted calories expended for years, through some mix of power data off the bike, heart-rate, and old-fashioned calculation based off calories consumed and weight change. I also spent six weeks this winter as a participant in a lab study that involved many test sessions on the bike measuring VO2 usage, providing another data point on my calorie expenditure. Now, I’m not as careful as (say) a professional bodybuilder would be about all this calorie counting, but the point is that I’m an experienced and accurate calorie counter drawing on a lot of experience and multiple methods of counting. Basically, my calorie counting is probably about as accurate as the average person will get, outside of a lab.

To quantify the metabolic change due to going vegan, I started by estimating the average number of calories I “burned” each active minute, per week, through the winter and spring. I work a sedentary job typing at a computer, so quantifying active minutes is pretty easy. My days break pretty cleanly into (a) sleep time, (b) time spent sitting at a computer, (c) time spent on casual walks, and (d) time spent in intensive workouts, either on the bike, running, or strength training. Active minutes include all time from (c) and (d).

WeekMinutes activeReal kcal demandkcal/active min4wk rolling avg
1/26102297113.4
2/0291319317.5
2/0959311225.6
2/1699285912.717.3
2/2394316816.718.1
3/0111324197.215.6
3/08105272710.711.8
3/15123306011.911.6
3/227917612.08.0
3/29107324315.410.0
4/05101270911.010.1
4/1212027379.59.5
4/19112294512.012.0
4/269425209.810.6
5/03*98297414.011.3
5/10*81306218.013.5
5/17*71253013.113.7
5/24*109278410.914.0
Metabolic data from the week ending 1/26, 2020, until the week ending 5/24, 2020. Weeks with * are weeks reflecting my new vegan diet. Active minutes is the daily average and includes both low intensity activity like walking, high intensity workouts on the bike, runs, and strength training. Real kcal demand is the actual number of daily calories required that week to maintain my weight, calculated by subtracting that week’s daily average body weight from the next week’s daily average body weight, then adjusting actual calories consumed up or down to account for that weight change. kcal per active minute is calculated by subtracting 1600 (my BMR) from real kcal demand and dividing the result by the number of active minutes.

To put these numbers in context, my best estimates are that I “burn” .5-1 kcal/min while sleeping, 1-1.5 kcal/min while awake and sitting, 6 kcal/min while walking or doing strength training, 10-12 kcal/min while doing moderate intensity aerobic exercise, 19 kcal/min working at my functional threshold power (FTP), and somewhere between 20-25kcal/min while working at my aerobic maximum.

Now, while some of the data is obviously distorted by the rough methods used to calculate it (e.g., the week ending 2/09 I could not have actually averaged 25.6 kcal “burned” per active minute), the overall picture looks reasonable and accurate, especially when measurement error is smoothed out via the 4wk rolling average. For example, averaging 17-18 kcal per active minute during February is reasonable. Since it was the dead of winter, I did very little walking, and that month was also the end of a long block of sustained aerobic workouts at or above FTP. After February, as it warmed up, I transitioned into a mix of casual walks and strength training, with bike workouts focused on either max aerobic capacity or all-out anaerobic sprint work. Just eyeballing the data, March and April look to average around 10.5 kcal per active minute, and that sounds right for this mix of activity.

Now, what’s interesting are the last four weeks of data, in May. Here I continued exactly the same workout pattern: lots of walks and workouts focused on strength training, aerobic max, and sprints. If anything, I’ve been doing more walks and fewer actual workouts. Still, my kcal per active minutes have increased dramatically. While the 4wk rolling average hovered right around that 10.5 for all of March and April, it shoots up to 14.0 in May.

Now, it seems unlikely that I’m actually “burning” nearly 40% more calories per active minute. That would be crazy. So, where are those extra calories coming from? Presumably they’re just coming from the rest of the day: while sitting, sleeping, etc. Now, I averaged 90 active minutes per day in those four weeks, and 90 x (14-10.5) is 315, meaning that I “burned” 315 extra calories per day while sitting, sleeping, etc. Again, it’s implausible that my actual energy demands went up. For example, sitting here typing still requires the same old 1-1.5 kcal/min it always did.

The plausible explanation is that the 315 “extra” calories needed are just undigested. I replaced the chicken, eggs, and cottage cheese I was eating as a source of protein with tofu, pea-based protein powder, and avocado. Since I also cut out a few other sources of animal fat (e.g., going to vegan friendly noodles), I’ve also been eating more vegetables to make up the lost calories. Presumably, vegetables, tofu (soy), pea extract, and avocado are harder to digest than chicken, eggs, cottage cheese, and animal fat. Many of the calories I now eat are just not being absorbed by my system, either being expelled as lost, undigested waste, or being absorbed by my gut microbiome.

To summarize, a preliminary analysis of the data suggests that in my first four weeks going vegan, I needed to consume somewhere around 300 kcals a day worth of food more than I was before to meet the same energy demands. That’s a shocking increase, over 11%. Whether it holds up long-term, or whether my body adapts in some way, is yet to be seen. I used to be a big believer that weight management was just a matter of balancing calories in vs calories out, but this is a pretty dramatic demonstration that since not all “calories in” are absorbed, just balancing this equation isn’t right. I like being wrong, and admitting as much.

Does this data support the idea that the average person could lose weight by keeping total calories consumed the same while going vegan? I think my case is too unlike the average person, since I went into this informal study with an already highly controlled diet and activity regime. That control makes my numbers fairly trustworthy, but also means its harder to extrapolate to conditions very unlike mine (i.e., a more typical lifestyle). Actual results, of course, are affected by many other factors, e.g. someone unintentionally and unknowingly increasing total calorie intake during the switch.