We have been taking near daily measurements through the period with some more occasional measurements of other parameters, so we have lots of data to misinterpret according to whatever bias or slant you might want to apply. As usual, life is complicated, and the data are subject to a lot of coarse- and fine-grained hypotheses that can be postulated to explain various features. We’ll offer an assortment of hypotheses, some of which are more strongly supported by the data than others. Since it’s all basically post-hoc analysis based on two subjects, none of the hypotheses can really be considered confirmed.
The most striking observation, perhaps, is that the change to lower-carbohydrate consumption continues to be working. (Let’s not call it a “diet” since that seems to mean something that people try to use as a temporary measure that ultimately fails when they revert to “normal” eating habits.) While our weight loss has inevitably slowed and had various “plateaus” and “reversals” or “setbacks,” our weight loss is holding and we are both at or near our lows of recent years some 30 months into our new eating habits. This bodes well for the long term. We are emphatically not counting calories, calorie restricting to the point of gnawing hunger, or otherwise depriving ourselves of the enjoyment of eating. Sure, we are emphasizing some different foods and limiting consumption of sugars of all sorts and simple starches, but, for the most part, we don't crave them and can satisfy what cravings we have with either small portions or satisfactory substitutes. Fortunately, we don’t often share meals with the high-carb/low fat crowd, so we don’t face a lot of peer pressure to “cheat,” and we stay away from the French bakeries. There are some differences between our preferred diets as well. David tends more towards the “Optimal Diet” (lower carb and high in butter and cream) while Cynthia gives in more to carb cravings (more fruit and indulgences such as Chinese dumplings), to which she attributes her various weight stalls and reversals!
And, of course, at least among those who are open-minded enough to have actually investigated the current state of the science, acceptance of the low-carb lifestyle has been increasing steadily. While advertising by the food industry continues to be abysmally misleading, we’ve noticed an increasing number of more positive references to low-carb nutrition in everything from a steadily increasing number of blogs posts, books, and scientific papers to passing references in recent movies (e.g., “She eats CARBS” from The Devil Wears Prada). The mainstream government agencies and medical societies are generally still not recognizing the error of the low fat diet, as seen in the 2010 USDA guidelines. (Quoting from Question 5 in Appendix E-1, “Conclusions”: “No optimal macronutrient proportion was identified for enhancing weight loss or weight maintenance. However, decreasing caloric intake led to increased weight loss and improved weight maintenance. Therefore, diets that are reduced in calories and have macronutrient proportions that are within the ranges recommended in the Dietary References Intakes (IOM, 2002/2005) (protein: 10%-35%; carbohydrate: 45%-65%; fat: 20%-35%) are appropriate for individuals who desire to lose weight or maintain weight loss. Diets that are less than 45 percent carbohydrate or more than 35 percent protein are difficult to adhere to, are not more effective than other calorie-controlled diets for weight loss and weight maintenance, and may pose health risk, and are therefore not recommended for weight loss or maintenance.”) There are, of course, huge entrenched economic interests that will continue to fight the status quo tooth and nail. There is some evidence that they may start to crack in the foreseeable future—for example, the American Diabetes Association now recognizes that a low carb diet may be useful for weight loss in diabetics—but progress continues to be slow.
Then there’s the exercise wild card. We all “know” that increased exercise is a “necessary” part of any “reputable” weight loss program. And yes, we have increased our level of exercise. We were never serious couch potatoes, but we weren’t serious athletes, either. Very roughly, we were running 20–30 mi/wk two and a half years ago, and increased to 30–40 mi/wk, and have often done closer to 50 mi/wk. We also started running ultramarathons of 30–50 mi in one day, averaging more than one such event per month at times. So what did all that exercise do for us? Well, we certainly got stronger and faster. We generally feel good and energetic (aside from the inevitable sore muscles and minor injuries). We also continue to find that more often than not, increased exercise correlates with weight gain, not weight loss! This happens both over the short term
(water weight of up to a few pounds the day after an ultra-marathon that may take a few days to lose), and over the longer term (our weight loss trends reversed for about 4 months after we started doing frequent day runs in excess of about 15 mi). We also tended to see a stalling of any downward trend whenever we increased our weekly mileage significantly. Partly, this is because moderate mileage increases such as this are easily compensated for by eating more. There are exceptions too, for example, during August 2009, Cynthia upped her mileage considerably (>50 mi/wk) and found she could not eat enough to keep her weight stable. However, this amount of mileage was not sustainable (due to an injury in May 2009 that began to cause pain), and eventually she gained it back.
There are, of course, several competing things going on when you exercise a lot. Over the period of the exercise itself, the dominant effect is usually level of hydration, and body weight is often used to monitor endurance athletes for dehydration and/or over-hydration. If you exercise hard enough and long enough, you can also deplete your glycogen stores to account for another pound or so of temporary weight loss (including the accompanying water of hydration). So generally speaking, you usually finish a long, hard bout of exercise down a few pounds. But then, of course, you eat and drink. Your appetite increases, so you may eat more than normal, and there are various reasons why you might retain extra fluid. That’s why we frequently saw a net increase in weight the day after. Interestingly, the size of this effect has generally decreased over time. Probably, as our bodies have become better adapted to the rigors of a long, hard day of exercise, they no longer see it as stressful. This is supported by the evidence of less muscle soreness and edema, as well. In David’s case, there is probably also a nutritional effect. He ran earlier events consuming more than normal carbs during and immediately after the event, and then, in later events, switched to more strict low-carb fueling before, during, and after. Some amount of fluid retention would be expected to be correlated with a temporary increase in carb consumption, and eliminating the carbs apparently eliminates that source of fluid retention. Cynthia tends to push harder into her non-aerobic zone, especially when trying to keep up with David, and chooses to consume more carbs during and after long runs, but she’s been showing smaller post-event weight spikes more recently, too. The effect is a sensitive measure of training, because it is more pronounced when less prepared or perhaps as a response to heat stress. The water retention is most likely due to a complex interplay of hormones that signal the kidneys to retain salt and fluid. Such a response is understandable after the stimulus of long and/or hot conditions.
The increased appetite can generally overcome any predicted weight loss from a purely thermodynamic point of view. The problem is that you have to run on the order of 30 mi or so to burn enough calories to consume a pound of body fat (assuming that you’re actually burning fat for fuel). If you do that over a week, it’s pretty easy to unknowingly increase your daily food Calorie consumption enough to more than offset that burn. It’s not much more than an extra couple of “servings” of something tasty per day. Be careful about rewarding yourself with too many bowls of ice cream or extra double cheesburgers!
An increased level of exercise, if done systematically and with adequate general nutritional support (enough protein, for example), often results in muscle building. This can result in body “recomposition”: loss of body fat and increase in lean muscle mass with no net change in weight. That may explain some of the apparent plateauing of our weights. Running doesn’t build bulky muscles the way, say, weightlifting does, but we have seen some measurable changes in body measurements.
Nevertheless, with a very blurry-eyed look at the weight loss data over 30 months, a simple-minded interpretation and hypothesis is that we both generally lost weight more or less linearly for 8 months until we started seriously increasing our level of exercise and then stalled out, remaining at a more or less constant weight for the following 22 months. (Click on figures to show larger.)
With slightly less blurry eyes, one immediately notices that longer trending period tends to follow more of an exponential curve rather than a straight line. A simple model which can be made to fit the data pretty well is to assume that you are always approaching an asymptote (target weight) exponentially so that your rate of weight loss (gain) gets steadily slower as you approach your target. Fitting such exponentials to the various regions on our graphs gives a pretty good fit (i.e., the data looks like it fits a set of straight line segments on a semi-log plot where an estimated target weight is subtracted out). Measured time constants vary from about half a year to two years. And while our weight loss is now hard to see from day to day or even week to week, we are still losing at an average net rate of about a pound every 2–3 months. We both feel like we should be able to lose another 10–15 pounds, but that could take a few years.
It is interesting to treat the data using some of the technical indicators typically used on financial charts. For example, one can draw upper("resistance") and lower ("support") levels and trend lines. Weight can bounce off or break through these lines as you can see on Cynthia's chart from August 2008 through February 2010. You can also see a downward trending channel or notice triangular patterns with converging oscillations, double bottoms, retracement levels, all very similar to observed price behavior on financial charts.
There are other secondary effects that may also be present in the data. While we have so far explained the weight gain last fall as due to increased exercise, it could also be due, at least in part, to a normal seasonal effect. Some weight gain through the fall and into mid-winter is perhaps genetically programmed to store fuel for the winter. Some of it may also be just increased fluid retention in cooler weather (or just reduced dehydration?—the body probably undergoes larger hydration cycles in hot weather as one sweats and eventually replaces lost fluid—but note that a drop in core body temperature actually has the reverse effect as anyone who dives in cold water can attest: the body naturally sheds excess water when cold).
Cynthia's data also show a pronounced oscillation with an amplitude of 4–6 pounds and a period of 1.5–3 months. We have, so far, been unable to correlate this oscillation with any obvious body cycles, lifestyle cycles, eating, or exercise habits. Being female, one might suspect menstrual cycle effects, but the period is too long and the amplitude is too large. (Menstrual cycles generally result in monthly weight variations with an amplitude of about 2 pounds. In order to see it, you typically have to average several months of data [with the end dates of cycles carefully lined up if the length of the cycle is at all irregular] since the amplitude is comparable to normal day to day fluctuations. It's actually more noticeable as a cyclic change in waist circumference.)
David's data show periods of unexpectedly rapid weight loss (July/August 2008, January/February 2009, May/June 2010). Again, we have not been able to clearly explain these periods, although similar “success” periods seem to be commonly reported anecdotally. Perhaps the body suddenly decides to adjust its natural setpoint in some important way. Fat storage and loss is driven more by hormonal signals than by daily calorie balance anyway.
If you want to keep losing weight, you may need to keep reducing your calorie consumption as well. In principle, this should happen automatically if you basically eat to satiety, but eating habits can often be somewhat independent of satiety if you are in the habit of eating particular portion sizes (e.g., 2 eggs and 2 slices of cheese, etc.). Presumably with a little conscious effort, you should be able to readjust your habits to your new needs as you lose weight, but some portions are a little hard to adjust. (It’s not convenient to cook 1.8 eggs for breakfast…)
Another interesting comparison is to plot David’s weight vs. Cynthia’s weight. This plot is noisier in that David’s and Cynthia’s weight gain and loss have not always been tightly correlated despite similar diet and exercise schedules. Overall, David’s weight is approximately 30% higher than Cynthia’s at any given time, but he has been losing about 1.2 pounds for each pound that Cynthia loses. We're still not sure how much more weight we can (or should) realistically lose. The corresponding weight loss rates are consistent with our college-age weights of about 167/125 pounds respectively, but a more realistic goal may be more like 175/133 pounds.
There are other measures of body composition that are often used to determine “ideal” weights. The most commonly used is the Body Mass Index or BMI. This is based purely on height and weight and does not take skeletal build or musculature into account at all. David is still classed as borderline “overweight” by standard BMI guidelines; Cynthia is “normal” at a BMI of ~23.
However, David is relatively well-muscled and big-boned. At least by current American on-the-street standards, most people would not say he was overweight at this point. Another approach to determining ideal body composition is to estimate percent fat. There are many ways to make this measurement—all approximations based on indirect measurements of one sort or another.
Underwater weighing is the current standard against which other measurements are typically evaluated, but it is imperfect, too. Skin-fold thickness is also popular, because it’s easy, but it can be unreliable. See Lyle McDonald’s post for more in-depth discussion. The two methods that are most readily available to most people (including us) are Bioelectric Impedance Analysis (BIA), a measurement built-in to some digital bathroom scales, and various formulas based on using additional body dimensions such as waist, hip, and neck circumference. Our favorite of these right now is a set of formulas derived by the US Navy based on height, waist, weight, neck, and hip (women only): %Fat=495/(1.0324 - 0.19077(log(waist - neck)) + 0.15456(log(height))) - 450 for men or %Fat=495/(1.29579 - 0.35004(log(waist + hip - neck)) + 0.22100(log(height))) - 450 for women. These give current values of 28.5% fat for Cynthia and 18.5% fat for David. Not surprisingly, these calculations put both of us solidly in the “acceptable” range, but still significantly above the upper end of the target ranges for athletes (presumably based on a young military test population: 20% for women, 13% for men). Just to give you some idea of the uncertainty in these measurements, the BIA method as implemented by a Weight Watchers bathroom scale gives 25.4% for Cynthia and 22.4% for David, showing discrepancies on the order of 3% and in opposite directions for Cynthia and for David.
And finally, just a quick observation about the Weight Watcher brand scale implementation of the BIA measurement: while we have been unable to locate any information on the algorithms implemented in the scale, it is clear that there is more than a little something amiss. First, it tends to report %Water in the mid-50s, while the human body is generally estimated to be closer to 70% water. While the instructions warn you that the data may be inaccurate if you take measurements immediately after heavy exercise or large fluid consumption, it is nevertheless disconcerting that it actually tends to report higher %Water (and lower %Fat) after losing a lot of fluid due to exercise and sweating, and lower %Water after a large drink! Clearly the algorithms and/or measurements fail to accurately account for variations in the distribution of fluid throughout the body. Another anomaly we have noticed is that the scale has reported a slight decrease in %Bone for both of us as we have lost weight, something that is very unlikely to be valid.
So all in all, our way of eating is pretty easy to sustain and requires no weighing or measuring. If we really wanted to lose weight faster, I'm sure it could be done using more discipline, but then we'd have to worry about regaining once the discipline slips. This way the changes are more gradual, and perhaps, more sustainable.
