Fruit is Bad for You (at Least it is for me)

Those of you who know me are aware of my love for data, and the fact that I collect all kinds of health indicators in a search for patterns.  Over the last year, I was shocked by the rise in my A1C levels, which is an indicator of pre-diabetes.  I was also aware of an occasional blood glucose level above 100 mg/dL, which is also associated with the onset of diabetes.  But, my blood glucose levels were not high enough to warrant A1C levels of about 5.7.  So I did what anyone of you would do, and that is to measure my daily blood glucose levels, giving better statistics than a weekly reading.

Indeed, I saw multiple weekly readings above 100 mg/dL with a high of 110 mg/dL.  Then I noticed in a particular week that not one reading was over 100 mg/dL.  The advantage of my highly structured diet is that it was simple to identify the change, and that was that my wife could not find fresh blueberries that week.  We kept a fresh bowel on the kitchen counter, and I would grab a fistful after a meal or as I was lured to them as I passed buy.  Being my favorite fruit, this my sole source of fructose, aside from a special low-sucrose and high-fat/protein chocolate that I make.  (Not to mention the occasional chocolate chip cookie that I have with my grandson when we visit Starbucks and the exquisite pretzels my sister-in-law ships in creative gift boxes every Christmas.)  Blueberries were the only source of regular fructose that I had been eating for the last couple years.

So, I remained off of blueberries for two months to get a decent amount of statistics, and was blown away by the results, shown in the plot.  First, this plot shows the importance of averaging.  There are large fluctuations in the test and in blood glucose levels that depend on all sorts of factors.  The blue open circles show the results while I was eating blueberries and the red solid circles during the time I stopped eating them.  The lines show a least squares fit to the data in the two regions.  The conclusion is that my blood glucose is a full 10 mg/dL lower when I'm not eating blueberries.  The beauty of averaging lots of data points is that the uncertainty can be determined from the scatter, and corresponds to about 1 mg/dL.  So one can argue whether the difference with and without blueberries is 10 mg/dL or 9 mg/dL, but the difference is both large and statistically significant.

I plan on continuing the zero fructose diet for another month or so, followed by another A1C test, which provides a blood glucose value averaged over three months.  This will allow the averages of both tests to be compared.

I will write more on this topic later when I have more time, but the conclusions are:

1. There is no such thing as a universally healthy diet.

2. There is no such thing as a healthy diet that is customized for a particular person.

Rather, our health is a struggle against internal and external forces, which adapts to evolutionary forces imperfectly.  The best we can do is chose a balance that makes each of us relatively healthy, and then we must accept the negative consequences.  In my case, a super-low-carb diet keeps me fit and trim, provides lots of energy and stamina, and has eliminated pesky annoyances such as heartburn and regular migraine headaches.  On the downside, my diet makes me prone to kidney stones.  These can be mitigated with simple meds, which are benign compared to treatments for diabetes and migraine headaches.  And physical stamina is more important to me than pleasing my sugar-craving reward center.

May your New Year be a happy one, optimized for adequate health and pleasures within reasonable constraints. 

Self Testing - Part I

 Those of you who know me will not be surprised that I love data.  Even noisy data can bring a picture into focus if there is enough of it.  I collect data on all sorts of things but find biometrics fun because it’s easy to collect and it can potentially give useful insights.

A couple of previous posts alluded to some obvious trends that I have observed.  To all those people who mocked me decades ago for starting a low carb diet, contrary to their predictions that I would gain weight, I shed 50 pounds by eating mostly fat with a bit of protein and as few carbs as possible.  In the process, my cholesterol also dropped to healthy levels (see http://unknownphysicist.blogspot.com/2011/10/eating-lots-of-fat-to-lose-weight.html).  I kept my weight down for a decade.

After a trip to Belgium and then to Italy, I adopted a Mediterranean diet, which was a gateway to me eating lots of carbs again, which produced an upward spiral in my weight.  Then in 2012, when I had almost regained all the weight that I had lost, I went on a strict low carb diet for a second time.  Again, it worked, but over the last two years, I have struggled with a small weight gain, whose source I have tried to identify using my vast stores of data.

In the interim, PSA (prostate specific antigen) data that I had been tracking took off, as I described in my post describing my ordeal with prostate cancer. (see http://unknownphysicist.blogspot.com/2021/01/can-major-surgery-cure-depression.html and http://unknownphysicist.blogspot.com/2020/06/i-was-relieved-i-had-cancer.html)

The graph below shows my weight as a function of time.  There are a series of drops each followed by a plateau.  Also plotted are my measurements of total cholesterol and LDL cholesterol along with error bars that reflect the accuracy of my instrument.  LDL is the bad cholesterol.  My triglycerides were always very low, so I was not so concerned about my cholesterol while it slowly fell over 4 years (1500 days).  Then, for no apparent reason, my cholesterol shot up even as I was losing more weight.

  

Both my doctor and I got concerned with the rise, so he prescribed the highest dose of atorvastatin – which scrubs serum cholesterol.  My cholesterol remained high for months while taking it, so he sent me to a cardiologist, who suggested I add ezetimibe, which prevents cholesterol from being absorbed through the intestines.  I was reluctant to take ezetimibe because I had had a bad experience with Vytorin, which combines atorvastatin with ezetimibe.  I hadn’t realized how much energy Vytorin was sopping away (and resulting in weight gain) until I accidentally stopped taking it for a week.  I then regained my usual vigor.  That’s when I stopped Vytorin and went back to my strict diet.

My cardiologist suggested that I take an ultralow dose of Ezetimibe, which I did at the time shown by the vertical gray line in the plot.  Indeed, it drastically decreased my serum cholesterol, as you can see in the plots, and I retained my energy levels within my ability to notice.  And my weight appeared to remain stable, as shown within the gray dashed box.

The plot below shows a magnified view of my weight after starting to take Ezetimibe.  The raw data (points) is quite noisy and can vary by as much as 6 pounds over a few days.  That variation is real.  There are days when I exercise vigorously and might also dehydrate, so I could easily lose a few pounds, which I regain when hydrating and remaining sedentary for a few days.

The raw weight data does not appear to show any trends.  However, a plot of 50 (dark curve) and 100 point smoothed data (gray curve) shows a different story.  My weight appears to be increasing and it also fluctuates over time.  The fluctuations might also be real due to seasonal variations in my diet and activity.  The gray vertical dashed line falls on the date that I started taking Ezetimibe and the pink dashed line when I had prostate surgery.  Each seems to have been responsible for some weight gain.

Next I fit the raw data to saturated exponentials starting at the time I began my Ezetimibe regimen (light blue curve) and then after prostate surgery (magenta curve).   The fit shows that the Ezetimibe resulted in almost a 3 pound weight gain (when averaged over a couple months).  This is consistent with the fact that I observed a huge weight increase when I was taking a larger dose.  I gained a little less than two pounds after my radical prostatectomy.  Though minor, I have noticed a little edema around the incision area, so this increase is also explainable.

The skeptic might complain that the pink curve start lower than the flat part of the blue curve.  A closer examination of the data right after my surgery shows that my weight dropped by 5 pounds from the previous baseline for a week right after surgery, so that biases the early parts of the fit.  However, all of the fits and the smoothed data nicely follow each other aside from seasonal variations.

I would not conclude form my data that ezetimibe results in weight gain nor that prostate surgery does so too.  However, the data is suggestive that this might be the case and my data is consistent with secondary observations.  I find it amazing how such small effects can be drawn from the data and gives a small level of confidence that it might be true.

In the meantime, I continue to take and analyze the data to see if any other correlations emerge that might signal an interesting underlying cause.  Apologies for typos and the like, but my Fitbit just informed me that I missed my 250 step goal this hour while writing.  So I have to get up and walk...