Blood Sugar Model Magik?

An interesting new-to-me study came on my radar this week “Personalized Nutrition by Prediction of Glycemic Responses” published by Zeevi et al in 2015. Now, if you’ve ever had the unfortunate experience of talking about food with me in real life, you probably know I am big on  quantifying things and particularly obsessed with blood sugar numbers. The blood sugar numbers thing started when I was pregnant with my son and got gestational diabetes. 4 months of sticking yourself with a needle a couple of times a day will do that to a person.

Given that a diagnosis of gestational diabetes is correlated with a much higher risk of an eventual Type 2 diabetes diagnosis, I’ve been pretty interested in what effects blood sugar numbers. One of those things is the post-prandial glucose response (PPGR) or basically how high your blood sugar numbers go after you eat a meal. Unsurprisingly, chronically high numbers after meals tend to correlate with overall elevated blood sugar and diabetes risk. To try and help people manage this response the glycemic index was created, which attempted to measure what an “average” glucose response to particular foods. This sounds pretty good, but the effects of using this as a basis for food choices in non-diabetics have been kind of mixed. While it appears that eating all high glycemic index foods (aka refined carbs) is bad, it’s not clear that parsing things out further is very helpful.

There are a lot of theories about why glycemic index may not work that well: measurement issues (it measures an area under a curve without taking in to account the height of the spike), the quantities of food eaten (watermelon has a high glycemic index, but it’s hard to eat too much of it calorie-wise), or the effects of mixing foods with each other (the values were determined by having people eat just one food at a time). Zeevi et al had yet another theory: maybe the problem was taking the “average” response. Given that averages can often hide important information about the population they’re describing, they wondered if individual variability was mucking about with the accuracy of the numbers.

To test this theory, they recruited 800 people, got a bunch of information about them, and hooked them up to a continuous glucose monitor and had them log what they ate. They discovered that while some foods caused a similar reaction in everyone (white bread for example), some foods actually produced really different responses (pizza or bananas for example). They then used factors like BMI, activity level, gut microbiome data to build a model that they hoped would predict who would react to what food.

To give this study some real teeth, they then took the model they built and applied it to 100 new study participants. This is really good because it means they tested if they overfit their model….i.e. tailored it too closely to the original group to get an exaggerated correlation number. They showed that their model worked just as well on the new group as the old group (r=.68 vs r=.70). To take it a step further, they recruited 26 more people, got their data, then feed them a diet predicted to be either “good” or “bad” for them.  They found overall that eating the “good” diet helped keep blood sugar in check as compared to just regular carbohydrate counting.

The Atlantic did a nice write up of the study here, but a few interesting/amusing things I wanted to note:

  1. Compliance was high Nutrition research has been plagued by self reporting bias and low compliance to various diets, but apparently that wasn’t a problem in this study. The researchers found that by emphasizing to people what the immediate benefit to them would be (a personalized list of “good” and “bad” foods, people got extremely motivated to be honest. Not sure how this could be used in other studies, but it was interesting.
  2. They were actually able to double blind the study Another chronic issue with nutrition research is the inability to blind people to what they’re eating. However, since people didn’t know what their “good” foods were, it actually was possible to do some of that for this study. For example, some people were shocked to find that their “good” diet had included ice cream or chocolate.
  3. Carbohydrates  and fat content were correlated with PPGR, but not at the same level for everyone At least for glucose issues, it turns out the role of macronutrients was more pronounced in some people than others. This has some interesting implications for broad nutrition recommendations.
  4. Further research confirmed the issues with glycemic index  In the Atlantic article, some glycemic index proponents were cranky because this study only compared itself to carb counting, not the glycemic index. Last year some Tufts researchers decided to focus just on the glycemic index response and found that inter-person variability was high enough that they didn’t recommend using it.
  5. The long term effects remain to be seen It’s good to note that the nutritional intervention portion of this study was just one week, so it’s not yet clear if this information will be helpful in the long run. On the one hand, it seems like personalized information could be really helpful to people…it’s probably easier to avoid cookies if you know you can still have ice cream. On the other hand, we don’t yet know how stable these numbers are. If you cut out cookies entirely but keep ice cream in your diet, will your body react to it the same way in two years?

That last question, along with “how does this work in the real world” is where the researchers are going next. They want to see if people getting personalized information are less likely to develop diabetes over the long term. I can really see this going either way. Will people get bored and revert to old eating patterns? Will they overdo it on foods they believe are “safe”? Or will finding out you can allow some junk food increase compliance and avoid disease? As you can imagine, they are having no trouble recruiting people. 4,000 people (in Israel) are already on their waiting list, begging to sign up for future studies. I’m sure we’ll hear more about this in the years to come.

Personally, I’m fascinated by the whole concept. I read about this study in Robb Wolf’s new book “Wired to Eat“, in which he proposes a way people can test their own tolerance for various carbohydrates at home. Essentially you follow a low to moderate carbohydrate paleo (no dairy, no legumes, no grain) plan for 30 days, then test your blood glucose response to a single source of carbohydrates every day for 7 days. I plan on doing this and will probably post the results here. Not sure what I’ll do with the results, but like I said, I’m a sucker for data experiments like this.

Data Driven Weight Loss: A Tale of Scales and Spreadsheets

In honor of the New Year and New Year’s Resolutions and such, I’m trying out a different type of post today.  This post isn’t  about statistical theory or stats in the news, but actually about how I personally use data in my daily life. If you’re not particularly interested in messy data, personal data, or weight loss, I’d skip this one.

Ah, it’s that time of year again! The first day of 2017, a time of new beginnings and resolutions we might give up by February. Huzzah!

I don’t mean to be snarky about New Year’s Resolutions. I’m actually a big fan of them, and tend to make them myself. I’m still trying to figure out some for 2017, but in the meantime I wanted to comment on the most common New Year’s Resolution for most people: health, fitness, and weight loss.  If Mr Nielsen there is to be believed, a third of Americans resolve to lose weight every year, and another third want to focus on staying fit and healthy. It’s a great goal, and one that is unfortunately challenging for many people. It seems there are a million systems to advise people on nutrition, exercise plans and other such things, and I am not about to add anything to that mix. What I do have however is my own little homegrown data project I’ve been tinkering with for the last 9 months or so. This is the daily system I use to help me work on my own health and fitness by using my own data to identify challenges and drive improvements. While it certainly isn’t everyone’s cup of tea, I’d gotten a few questions about it IRL, so I thought I’d put it up for anyone who was interested in either losing weight or just seeing the process.

First, some personal background: Almost exactly 2 years ago (literally: December 31st, 2014), I decided to start meeting with a nutritionist to help me figure out my diet and lose some weight. Like lots of people who have a lot on their plate (pun intended), I had a ridiculous amount of trouble keeping my weight in a healthy range. The nutritionist helped quite a bit and I made some good progress (and lost half the weight I wanted to!), but I realized at some point I would have to learn how to manage things on my own.  Having an actual person track what you are doing and hold you accountable is great and was working well, but I wanted something I could keep up without having to make an appointment.

Now, the math background: Around the same time I was pondering my weight loss/nutritionist dilemma  I got asked to give a talk at a conference on the topic “What Gets Measured Gets Managed”. One of the conference organizers had worked with me a few years earlier and said “I know you were always finding ways of pulling data to fix interesting problems, do you have anything recent you’d like to present?” Now this got me thinking about my weight. How was it that I could always find a data driven way to address a work problem, but couldn’t quite pull it together for something important to me in my personal life? I had tried calorie counting in the past, and I had always gotten frustrated with the time it took and the difficulty in obtaining precise measurements, but what if I could come up with some simpler alternative metrics?  With my nutritionists blessing (she had a remarkable tolerance for my love of stats), I decided to work on it.

The General Idea: Since calories were out, I decided to  play around with the idea of giving myself a general “score” for a day. If I could someone capture a broad range of behaviors that contributed to weight gain and the frequency in which I engaged in them, I figured I could figure out exactly what my trouble spots were, troubleshoot more effectively and make sure I stayed on track.  At the end of each week I’d add up my weekly score and weigh myself. If I lost weight, no problem. If I gained weight, I’d tweak things.

The Categories: The first step was to come up with categories that covered every possible part of my day or decision I felt contributed noticeably to my weight. I aimed for 10 because base 10 rules our lives. My categories fell in to four types:

  1. Meals and snacks I eat 3 meals and 2 snacks each day, so each got their own category.
  2. Treat foods Foods I need to watch: sweets/desserts, chips, and alcohol each got their own category
  3. Health specific issues I have celiac disease and have to avoid gluten. Since eating gluten seems to make me either ridiculously sick or ravenously hungry, I gave it a category so I could note if I thought I got exposed
  4. Healthy behaviors I ultimately only track exercise here, but I have considered adding sleep or other non-food behaviors too.

The Scores:  Each score ranges from 0 to 5, with zero meaning “perfect, wouldn’t change a thing” and five meaning “gosh, that was terribly ill-advised”.  Between those two extremes, I came up with a slightly different scoring system for each category.

  1. Meals and snacks Basically how full I feel after I eat.  I lay out a reasonable serving or meal beforehand, and then index the score from there. If I take an extra bite or two because the food just tastes good, I give myself a 1. If I was totally stuffed, it’s a 5. Occasionally I’ve even changed my ranking after the fact when I get to the next meal and discover I’m not hungry.
  2. Treat foods One serving = 1 point, 2 servings = 3 points, more than that is 4 or 5. The key here is serving. Eating a bunch of tortilla chips before a meal at a mexican restaurant is almost never one serving, and a margarita at the same restaurant is probably both alcohol and sugar. It helps to research serving sizes for junk food before attempting this one.
  3. Health specific issues For gluten, if I think I got a minor exposure, it’s a 1. The larger the exposure I got, the higher the ranking I give it. The day I got served a hamburger with what was supposed to be a gluten free only to discover it wasn’t? Yeah, that’s a 5.
  4. Exercise I generally map out my workouts for the week, then my score is based on how much I complete. A zero means I did the whole thing, a 5 means I totally skipped it. I like this because it incentivizes me to start a workout, even if I don’t finish it.

With 10 categories ranked 0 to 5, my daily score would be somewhere between 0 “perfect wouldn’t change a thing” and 50 or “trying to re-enact the gluttony scene from Seven“.  To start, I figured I’d want to be below a score of 5 per day or 35 per week. Since I am not built for suffering, that seemed manageable.

Obviously all of those scores are a bit of a judgment call. If I lose track of what I ate or feel unhappy with it, I give myself a 5. I try not to over think these rankings to much, and just go with my gut. That mexican meal with the chips and margarita for example was a 5 for the chips, 3 for feeling full after dinner, 1 for the alcohol and 1 for the sugary margarita. Is that 100% accurate? Doubtful, but does a score of 10 seem about right for that meal? Sure. Will my scale be lower the day after a meal like that? No. A score of 10 works. With the categories and the score, my weekly spreadsheets end up looking like this:


How I use this data: Okay, so remember how I started this with “what gets measured gets managed”? I use this data to find weak spots, figure out where I’m having the most trouble, and to come up with solutions. For example, every month I add my scores up and figure out which category is my worst one. When I first started, I realized that I actually skipped a lot of workouts. When I looked at the data, I noticed that I would have one good week of working out followed by one bad week. When I thought about it, I realized I was trying to complete really intense workouts, and that I was basically burning myself out and needing to take a week off to regroup. When I decided to actually decrease the intensity of my workout plan, I stopped skipping days. Since the workout you actually do tends to be better than the one you only aspire to do, this was a win. Another trend I noted was that I frequently overate at dinner. This was solved by packing a bigger lunch. There’s a few other realizations like this, and they all had pretty simple fixes. For January I’m working on reducing the number of chips I eat, because damn can I not eat chips in moderation.

The results: So has this worked? Yes! Since April, I’ve lost almost 5 points off my BMI, which takes me from the obese/overweight line to the healthy/overweight line. Here’s my 7 day moving average (last 7 days averaged together) score plotted against a once a week weigh in. The red line is my goal of 5:


Note: There are some serious jumps on this chart, mostly because I can retain a crazy amount of water if I eat too much sodium.

At the moment I’ve decided to give myself a month off from weigh-ins since the holidays can be so crazy, but on that last weigh in I was only 3 lbs away from being in the normal BMI range. As I mentioned, I’m not built for suffering. Slow weight loss is fine with me.

It’s interesting to note that I actually don’t make my goal of 5 or less per day all the time. Over the 274 days I’ve been tracking, I only was at 5 or under about 70% of the time. I still lost weight. I’ve thought about raising my limit and trying to stay under it all the time, but as long as this is still working I’m going to stick with it.

General thoughts: Much of the philosophy behind how I pulled this data actually comes from the quality improvement “good enough” world, as opposed to the hard research “statistical significance” world. The weigh in data is always there to test my hypotheses. If my scoring system said I was fine but my weight was going up, I would change it. I’m sure that I have not accurately categorized every day I’ve had since April, but as long as my daily scores are close enough to reality, it works. It’s the general trend of healthy behaviors that matters, not any individual day. The most important information I’ve gotten out of this process is what small tweaks I can make to help myself be more healthy. Troubleshooting the life I actually have a getting specific feedback about which areas I have problems with has been immensely helpful. Too often health and diet advice advises us to impose Draconian limits on ourselves that set us up for failure. By tracking specific behaviors and tweaks over the course of months, it’s a lot easier to figure out the high impact changes we can make.

If I had any advice for anyone wanting to try a similar system, it would be to really customize the categories you track and to think through a ranking system that makes sense to you. Once I invented my system, I actually only have to spend about 45 seconds a day ranking myself. I only change things if I see the weight creeping up or if some piece seems to not be working. At this point I review the categories and scores monthly to see if any new patterns are emerging. In the quality improvement world, we call this a PDSA cycle: Plan, Do, Study, Act. Plan what you want to do, do what you said you would do, study what you did, act on the new knowledge. By having data on individual aspects of my daily life, this process became more manageable.

Happy tracking!