Category Archives: bad science

Is Paula Deen’s Food Less Healthy? Intrepid NYTimes Journalist Finds Out!

Or not really, but publishes the article anyway!

Glenn Collins of the NYTimes asks:

After the hue and cry following Paula Deen’s announcement that she has Type 2 diabetes and would become a paid spokeswoman for a drug company, a question nags: Is the food that emerges from her kitchen really less healthy than the cuisine from other restaurants?

So he had a meal from Deen’s restaurant (fried chicken, collard greens, and macaroni and cheese, which he calls “reasonably sized”) and a dish from an Italian restaurant in Greenwich Village (sausage and peppers with polenta, which he calls “hardly cuisine minceur”) sent to Certified Laboratories in Plainview, NY for nutritional analysis.

image  imageimage

On the one hand, he seems to realize how pointless this is, admitting that the comparison was “unscientific and arbitrary.” Turns out the fried chicken meal has more than 2x as many calories and grams of fat, but less than a third as much “sugar” (unclear what that measures—sucrose? glucose? fructose? lactose? all of the above?) of the sausage and polenta. No word on total carbs. Is there a lesson to be drawn from this “subjective test,” he asks? (Which is also wrong: the problem is not that it’s “subjective”—the nutritional analysis is objective, all right. The problem is that it’s completely meaningless because he made no attempt to control any of the variables.)

Well, lovers of both Deen and Frankies are hereby advised to consider moderation and not look to Ms. Deen for enlightenment.

Many things about this article confuse me. Why not compare a meal from Lady & Sons to a comparable meal at another restaurant: i.e. find another fried chicken, mac & chs, and collard greens plate? Why not do an analysis of the nutritional breakdown of several of her recipes compared to other people’s versions? Or try to figure out the average nutritional content of a Lady & Sons meal and compare it to FDA recommendations? All of those methods would still have limitations, but at least you’d be comparing oranges and oranges.

I think the most head-scratching word in the entire article is the “hereby” in the above quote.

What information presented here makes the case for moderation—and moderation in what? The sausage and polenta dish has 347 calories and 15 grams of fat, which is probably less of both of those things than I eat for breakfast most days. And Deen’s fried chicken dinner comes in at only 830 calories, which is less than most of Subway’s $5 Footlongs (tuna salad on 9-grain wheat without cheese or extra sauce has a total of 930 calories; turkey breast on Italian with regular mayonnaise and no cheese has 860 calories; the meatball marinara on Italian Herbs & Cheese with no extra cheese or sauce has 1050). Neither of them seem particularly excessive to me; in fact, I’m pretty sure eating just the polenta dish for dinner would leave me feeling hungry. Perhaps Frankie’s is the kind of place where most people order multiple courses?

Has anyone has ever turned to Paula Deen for medical or nutritional advice? As even she noted when she made the announcement that she has Type 2 Diabetes on the Today Show, she’s a cook, not a doctor. I’m not impressed with her decision to use her diagnosis as an opportunity to profit by acting as a spokesperson for pharmaceutical company Novo Nordisk, who make the diabetes drug Victoza she claims to be taking. Fortunately, based on some preliminary reactions, most ad executives and consumers aren’t loving it either. And I’m really not sure what information in this article is supposed to have any bearing on Deen as a source of “enlightenment.” Actually, she might actually be able to provide some great insight about building a food brand, being dissed by Anthony Bourdain, partnering with ethically suspect corporations, getting them to donate food to charity, and mugging terrifyingly for the camera.

image image image  image

At least the article refrains from blaming Deen for her diabetes, the way so many others have in the wake of her announcement. As if anyone knows how she eats on a day to day basis anyhow or the relationship between diet, weight, and diabetes was simple and universally agreed-upon. But I understand the tendency to moralize about health. In some ways, this article is much more mysterious. If you’re going to go to the trouble and cost of having nutritional analysis done to try to figure out if Paula Deen’s food is really “worse” than anyone else’s, why not do it in a way that would actually enable you to reach something approximating a conclusion?

Why Posting Calorie Counts Will Fail, Part III: Calorie-restriction dieting doesn’t work long-term for most people

Previously in this series: Intro, Part I, and Part II.

The article on "Making Weight Loss Stick" is by Bob Greene, the personal trainer and "fitness guru" Oprah first started consulting with in 1996. Sadly, I don't think that's *meant* to be ironic. Oprah 2005/2009

To test whether turning [fat people] into thin people actually improves their health, or is instead the equivalent of giving bald men hair implants, it would be necessary to take a statistically significant group of fat people, make them thin, and then keep them thin for long enough to see whether or not their overall health then mirrored that of people who were physiologically inclined to be thin. No one has ever successfully conducted such a study, for a very simple reason: No one knows how to turn fat people into thin people.
Paul Campos, The Obesity Myth (2004)

Diets do cause weight loss…in the short term

People who think calorie restriction dieting “works” haven’t necessarily been duped by the diet industry or seduced by the prevailing nutritional “common sense” that weight loss and gain are a simple matter of calories in vs. calories out. Many of them believe it because their personal experience seems to confirm it, often repeatedly. Of course, “repeatedly” is part of the problem. Weight cycling—losing and re-gaining 5% or more of one’s total body weight—isn’t what dieters or public health policy makers are shooting for. Even people dieting with a specific occasion in mind, like a wedding or a high school reunion, would generally prefer to achieve permanent weight-loss.

But almost a century of research has shown that dieting—which usually involves calorie restriction—is not the way to do that. Studies repeatedly find that while eating less causes weight-loss in the short term, a majority of participants in weight-loss interventions focused on diet gain most of the weight back within 1 year and the vast majority (90-95%) gain all of it back within 3-5 years. Approximately 30% gain back more than they initially lost, and there’s some evidence that people who’ve lost and regained weight have more health problems than people who weigh the same, but have never lost and regained a significant amount of weight.

This is not controversial. Virtually every study of weight-loss dieting that has followed participants for longer than 6 months has found that the majority of dieters regain all the weight they lose initially, if not more. In other words, Oprah’s high-profile weight fluctuations are not the unfortunate exception to most dieters’ experience, they are the rule. A gallery of pictures of Oprah through the years illustrates the most frequent and reliable outcome of dieting:

Oprah in The Color Purple Screen shot of the infamous "fat wagon" episode first aired in the fall of 1988, when Oprah strode on set in what she proudly declared were size 10 Calvin Klein jeans after an Optifast diet, wheeling a Red Flyer wagon full of lard representing how much weight she'd lost  At the Emmy Awards, holding her third for "Outstanding Talk/Service Show Host"  Holding yet another Emmy at the end of that impressively-muscled arm, shaped with the help of trainer Bob Greene

             1985                           1988                             1992                             1996

 At the party celebrating the first anniversary of O Magazine  At the Academy Awards, wearing Vera Wang Presenting at the Emmy Awards presenting at the 2010 Oscars, possibly on the way back down again?

              2001                            2005                          2008                            2010        

I am not concerned (in this entry) with why calorie restriction diets fail—there are competing theories and perhaps I’ll try to tackle them some other time. However, when it comes to evaluating public health policies aimed at the general population, like posting calorie counts on menus, it doesn’t really matter why the kind of behavior it’s designed to encourage fails, especially when it fails so spectacularly. Whether the problem is that 90-95% of people don’t have the willpower to stick to calorie-restricted diets or that most peoples’ metabolic rates eventually adjust or both or something else entirely, continuing to prescribe calorie restriction to individuals seeking to lose weight is futile at best. Given the health problems associated with weight cycling and psychological distress caused by diet “failure,” it’s probably also dangerous and cruel. More on that another day, too.

The goal of this entry is to provide a condensed-but-comprehensive overview of the evidence that convinced me that weight-loss dieting—and particularly calorie-restriction dieting or eating less—does not “work” for most people. By “work” I mean lead to significant weight loss—at least 10% of starting body weight—that lasts for more than 3 years (in keeping with the clinical definition of “weight loss success” proposed by the 1998 National Heart, Lung, and Blood Institute [NHLBI] Obesity Education Initiative Expert Panel proposed). I honestly tried to keep this as short as possible and bolded the “highlights” if you want to skim. However, if brevity is what you’re looking for, see this 2007 Slate article.

A Meta-Review of the Literature

Of course, I’m not the first person to try to figure out what kind of picture decades of weight-loss research was painting. I found 14 reviews of weight-loss research in peer-reviewed journals (Mann et al 2007, Jeffrey et al 2000, Perri & Fuller 1995, Garner & Wooley 1991, Jeffrey 1987, Bennett 1986, Brownell & Wadden 1986, Brownell 1982, Foreyt et al 1981, Wilson & Brownell 1980, Stunkard & Penick 1979, Wooley et al 1979, Foreyt 1977, Stunkard & Mahoney 1976). And they all say basically the same thing: calorie-restriction diets don’t work long-term. Here’s how three of the most recent ones came to that conclusion, and one meta-analysis that claims to challenge the consensus, although it turns out that all they’ve really done is redefine “success.” 

Diets Don’t Work—Mann et al 2007 (free full text): This review of 31 weight-loss studies by a team of UCLA researchers was aimed at developing recommendations for Medicare regarding obesity prevention and treatment. They were only able to find 7 studies of weight-loss dieting that randomly assigned participants to diet or control groups and followed them for at least two years (the “gold standard” required to make causal claims about the effects of dieting). And the “gold standard” studies did not support the claim that dieting promotes significant or long-term weight loss:

Across these studies, there is not strong evidence for the efficacy of diets in leading to long-term weight loss. In two of the studies, there was not a significant difference between the amount of weight loss maintained by participants assigned to the diet conditions and those assigned to the control conditions. In the three studies that did find significant differences, the differences were quite small. The amount of weight loss maintained in the diet conditions of these studies averaged 1.1 kg (2.4 lb), ranging from a 4.7-kg (10.4-lb) loss to a 1.6-kg (3.5-lb) gain. (223)

They also examined 14 studies with long-term follow-ups that didn’t involve control groups. The average initial weight loss in those studies was 14 kg (30.8 lb), but in the long-term follow-ups, participants typically gained back all but 3 kg (6.6 lb). Of the eight studies that tracked how many participants weighed more at the follow-up than before they went on the diet, the average was 41% with a range of 29%-64%, and in every case was higher than the percentage of participants who maintained weight loss. In other words, participants were more likely to regain more weight than they initially lost than they were to maintain their initial weight loss. Although Mann et al note several problems with these studies, like low participation rates in the long-term follow-ups, heavy reliance on self-reporting as the primary or only measure of weight, and failure to control for the likelihood that some of participants were already dieting again at the follow-up, those factors should have biased the results in the direction of showing greater weight-loss and better long-term maintenance, not less.

Finally, they looked at 10 long-term studies that didn’t assign participants to “diet” or “non diet” conditions randomly. In general, these were observational studies that assessed dieting behavior and weight at a baseline time and then followed up with participants to measure changes in behavior and weight over time. Of those studies, only 1 found that that dieting at the baseline led to weight loss over time, 2 showed no relationship between dieting at the baseline and weight gain, and 7 showed that dieting at the baseline led to weight gain.

Their recommendation to Medicare:

In the studies reviewed here, dieters were not able to maintain their weight losses in the long term, and there was not consistent evidence that the diets resulted in significant improvements in their health. In the few cases in which health benefits were shown, it could not be demonstrated that they resulted from dieting, rather than exercise, medication use, or other lifestyle changes. It appears that dieters who manage to sustain a weight loss are the rare exception, rather than the rule. Dieters who gain back more weight than they lost may very well be the norm, rather than an unlucky minority. If Medicare is to fund an obesity treatment, it must lead to sustained improvements in weight and health for the majority of individuals. It seems clear to us that dieting does not. (230)

Long-term Maintenance of Weight Loss: Current Status—Jeffrey et al 2000 (free abstract or full text with umich login): A review of 20 years of long-term weight loss studies that describes the weight loss and regain among patients who participate in behavioral treatments for obesity as “remarkably consistent” (7) which is visually represented by lots of graphs of different studies on the long-term results of weight loss studies that all pretty much look the same:

Very low calorie diets vs. Low calorie diets (Wadden et al 1993)  Fat restriction vs. calorie restriction (Jeffrey et al 1995)

Diet only vs. Diet + exercise (Sikand et al 1988) People who were paid $25/wk for successful weight loss vs. people who weren't paid (Jeffrey et al 1993)

Basically no matter what researchers do, most dieters achieve their maximum weight loss at 6 months and then gradually regain all or almost all of the initial weight lost within 3-5 years, if not faster. They conclude:

The experience of people trying to control their weight is a continuing source of fascination and frustration for behavioral researchers. Overweight people readily initiate weight control efforts and, with professional assistance, are quite
able to persist, and lose weight, for several months. They also experience positive outcomes in medical, psychological, and social domains (NHLBI Obesity Education Initiative Expert Panel, 1998). Nevertheless, they almost always fail to maintain the behavior changes that brought them these positive results. Moreover, as we hope we have shown, efforts to date to change this weight loss-regain scenario have not been very successful.

Confronting the Failure of Behavioral and Dietary Treatments for Obesity—Garner and Wooley 1991 (free abstract or full text with umich login): Like Mann et al, Garner and Wooley were seeking to translate the available evidence about weight-loss dieting into recommendations for treatment—in this case, best practices for mental health practitioners seeking to counsel and treat overweight and obese patients. They note that short-term behavioral studies consistently show that modifications in eating and exercise behaviors lead to weight-loss, but that as the duration of studies increases, “over and over again the initial encouraging findings are eroded with time” (734).

The authors are particularly perturbed that poor results are often presented by study authors as positive. For example, an 1981 study comparing standard behavioral therapy with a weight-loss drug, or the therapy and drug combined found that all of the treatment groups lost a significant amount of weight in the first 6 months, and then all of the treatment groups showed significant re-gain by the end of the 18 month follow-up.the consistency in the curves is really eerie after a while...the 6 month nadir, the gradual incline; also, it is completely baffling to me how someone could look at this graph and think the most notable part is the gap between the three treatments at 18 months

Instead of concluding that all of the treatments had failed to produce lasting weight loss, the authors conclude that these results provide hope for behavioral therapy, because that group showed the slowest rate of weight re-gain:

This most recent study provides grounds for optimism as to the future of behavioral treatment of obesity . over the long run, behavior therapy clearly outperformed the most potent alternative treatment with which it has yet been compared. (734 in Garner and Wooley, 135 in the original)

This pattern is nearly as consistent as the finding that weight is gradually regained and many individuals eventually weigh more than they did at the start of the treatment. After four years, nearly all participants in nearly all studies gain back nearly all the weight they initially lost: Adams, Grady, Lund, Mukaida, & Wolk, 1983; Dubbert & Wilson,1984; Kirschenbaum, Stalonas, Zastowny, & Tomarken, 1985; Murphy, Bruce, & Williamson, 1985; Rosenthal, Allen, & Winter, 1980, Bjorvell & Rossner, 1985; Graham, Taylor, Hovell, & Siegel, 1983; Jordan, Canavan, & Steer, 1985; Kramer, Jeffery, Forster, & Snell, 1989; Murphy et al. 1985; Stalonas, Perri, & Kerzner, 1984; Stunkard & Penick, 1979. And yet, the authors of those studies insist that the diet interventions are “effective,” sometimes claiming that if the subjects had not dieted they would weigh even more. They almost never admit that the treatments completely failed to do what they set out to do, which is produce a clinically significant weight loss that can be maintained long-term. When they do admit that the results are “disappointing,” they frequently call for more “aggressive” treatments like very low calorie diets (VLCD or <800 kcal/day) or supervised fasting (which is no longer approved because of the risk of mortality).

Garner and Wooley also evaluate studies that used VLCD, some of which involved Optifast, the protein shake that Oprah used to achieve her 67 lb weight loss in 1988. Just like with other calorie-restriction diets, people on VLCD generally lose weight initially, although drop-out rates are much higher than in other weight loss studies (50% or more). Participants who stick to the diet typically maintain the weight loss for about a year, but regain most if not all of the weight they lost within three years and many gain more than they initially lost (Swanson and Dinello, 1970, Sohar and Sneh, 1973, Stunkard and Penick, 1979, Johnson and Drenick 1977, Drenick SC Johnson, 1980, Wadden et al., 1983, Wadden, Stunkard, & Liebschutz 1988, Hovel et al., 1988). Based on all of those studies, they conclude:

Although the rate and magnitude of weight loss have been the basis for recommending the VLCD, its most remarkable feature is the speed of weight regain following treatment. (740)

Garner and Wooley found only two studies of weight-loss dieting that reported better long-term results, and both had extremely low rates of participation in the follow-up and relied on self-reported weights. For example, Grinker et al (1985) reported that 55% of the participants in a residential treatment program had maintained a 5-kg weight loss based on the responses of only 38% of the original participants. They suggest that it seems far more likely that the low participation in the follow-up biased the results than that those studies are right and all the other ones or wrong and conclude:

It is only the rate of weight regain, not the fact of weight regain, that appears open to debate. While this may be discouraging to the individual intent on weight loss, it should also provide some solace to the many individuals who have failed at dieting and have attributed the failure to a personal lack of will power. (740)

It is difficult to find any scientific justification for the continued use of dietary treatments of obesity. Regardless of the specific techniques used, most participants regain the weight lost. (767)

They make the following recommendation to mental health practitioners:

We suggest that at the least, if weight loss is offered, it should be done with full disclosure of the lack of long-term efficacy and of the possible health risks [which, as they explain, include physical and psychological risks correlated with weight fluctuation]. It is further recommended that alternative nondieting approaches aimed at improving the physical and psychological well-being of the obese individual be given priority over dietary treatments as a subject of research and that such treatments be offered on an experimental basis. (767)

Long-term weight-loss maintenance: a meta-analysis of US studies—Anderson et al 2001 (free full text): As the title suggests, this is a meta-analysis rather than a review article, meaning rather than summarizing and evaluating what other studies found, they lumped together the data from 29 different studies. 13 of the studies involved “very low energy diets” (VELDs), 14 involved “hypoenergetic balanced diets” (HBDs) and 2 involved both—in other words, they were all calorie-restriction diets, and about half of them required participants to eat less than 800 kcal/day. The authors claim that no long-term randomized, controlled studies were available, and it’s unclear why they didn’t think studies like Jeffrey and Wing 1995 (discussed below) should count.

They don’t provide details for any of the studies individually, but do disclose that the number of participants ranged from 6 to 504, the length of treatment ranged from 8 to 30 weeks, average initial weight loss ranged from 3.5 to 37.9 kg for women and 6.2 to 44.2 kg for men, and follow-up participation rates ranged from 50% to 100% with a median of 82%. In other words, these were very different studies. Here are the results of their aggregation of the data:

again, what they're focusing on is the relatively small loss maintained by year 5 rather than, say, the precipitous drop from year 1 to year 2

The average weight loss at 5 years for both VELDs and HBDs was 3.0 kg, or ~3.2% of the participants’ starting weight and 23.4% of their initial weight loss. Anderson et al conclude:

These average values are higher than those reported in earlier studies and indicate that most individuals who participate in structured weight-loss programs in the United States of the type reported in the literature do not regain all of the weight lost at 5 y. of follow-up.

Sure, not all of the weight, only 76.6% of it. It still seems to me like a perversion of the idea of “success” to claim that these results show that calorie-restriction diets are “effective.” The average initial weight loss was 14 kg. If you lost almost 31 lbs and then regained 25 lbs, would you consider your diet a long-term success? Mann et al wouldn’t. In the 14 long-term studies without control groups that Mann et al evaluated, they also note an average maintenance of ~3 kg. They just don’t think that’s very impressive:

It is hard to call these obesity treatments effective when participants maintain such a small weight loss. Clearly, these participants remain obese. (Mann et al 223)

Interpretation/equivocation aside, there’s still some discrepancies between their analysis and the consensus in the other reviews which I wish I could explain. It’s not like this was a study of a new treatment—they relied exclusively on existing studies, at least some of which were also included in the reviews of the literature discussed above. However, some of the studies they included must have reported (possibly significantly) better results to bring up the average. Since they didn’t evaluate the studies individually, it’s impossible to tell from their write-up whether those studies involved some sort of strategy that made calorie restriction dieting “work” (and somehow didn’t attract widespread attention) or whether the results in those studies were biased by low participation rates in follow-ups, self-reporting, or some other factor(s).

A Closer Look at the Studies Themselves

I have not read every single study referenced in the review articles, although I have at least glanced at many of them. The ones I chose to explore in further depth here either 1) meet the “gold standard” of randomized assignment to diet/non-diet conditions and at least 2 years of follow-up or 2) are too recent to be included in the review articles.

Long-term Effects of Interventions for Weight Loss—Jeffrey and Wing 1995 (free abstract or full text with umich login): This is one of the seven studies included in the first part of the Mann review. 202 participants between the ages of 25 and 45 who were between 14-32 kg above the MetLife standards for the “ideal weight” for their height were randomly assigned to one of five experimental groups:

  • a control group which received no intervention
  • a standard behavioral therapy group (SBT) that received instruction on diet (including advice on how to follow a 1000-1500 calorie/day diet), exercise (including the recommendation to walk or bike 5 days/wk with an initial goal of burning 250 kcal/wk and gradually increasing that to 1000 kcal/wk), and behavior modification (including keeping food and exercise diaries. This advice was given in weekly counseling sessions for the first 20 weeks and monthly sessions thereafter for a period of 18 months.
  • a SBT + food group, which received the same counseling along with premeasured and prepackaged breakfasts and dinners for 5 days/week for 18 months
  • a SBT + $ incentive group, which received the same counseling along with up to $25/week  for achieving and maintaining weight loss
  • a SBT + food + $ incentive group, which got the counseling, meals, and money

In addition to the 18 months of the study, the participants were contacted at 30 months (a full year after the study ended) for an additional follow-up, which was completed by 177 (88%) of the original participants. Here are the results:

 is this shape getting familiar? 

All the treatment groups lost weight during the intervention, achieving their maximum results at 6 months. However, by 12 months—even though they were all still receiving the treatment, they were beginning to regain weight. By 30 months, there was no significant difference between any of the treatment groups and the control group. The authors wheedle a bit, claiming the difference “approaches levels of statistical significance” (.08), but are honest enough to admit in the end:

The overall results of this evaluation reemphasize the important point that maintaining weight loss in obese patients is a difficult and persistent problem.

Preventing Weight Gain in Adults: The Pound of Prevention Study—Jeffrey & French 1999 (free full text): This more of a “failure of low-cost educational interventions designed to encourage weight loss” than a failure of weight loss dieting per se, but it’s still relevant because 1) the experimental group “got the message” communicated in the educational intervention but gained the same amount of weight over 3 years as the control group and 2) calorie labeling is essentially a large-scale, low-cost educational intervention. The idea that education will make people thinner relies on the assumption that people would not be (as) obese if they only knew they were gaining weight, that they should eat more fruits and vegetables, that they should reduce their consumption of high-fat foods, and/or that they should get more exercise.

But most people do know all those things. In this study, 228 men and 594 women employed by the University of Minnesota and 404 low-income women, all between the ages of 20-45, were recruited to participate in a 3-year study. Half of the participants were assigned randomly to a control group and the other half were assigned to the “intervention” group, which received a 2-4 pg monthly newsletter called Pound of Prevention. The newsletter emphasized five themes:

1) weighing yourself regularly (at least once a week)
2) eating at least 2 servings of fruit per day
3) eating at least 3 servings of vegetables per day
4) reducing the consumption of high-fat foods
5) increasing exercise, especially walking

In other words, “common sense” nutritional advice, although not explicitly calorie reduction. The newsletter included recipes, suggested particular areas/routes in the local areas for walking, and included a return-addressed, stamped postcard asking participants to report their current weight and also answer whether they had walked for 20 minutes or more, eaten 2 servings of fruit, eaten 3 servings of vegetables, or weighed themselves in the last 24 hours. Intervention participants were also invited to take part in a variety of activities during the three years, including 4-session weight control classes, aerobic dance classes, free 1-month memberships to community exercise facilities, walking groups, and a walking competition. Additionally, half of the “intervention” group was assigned randomly to an “incentive” group who were eligible for a monthly $100 lottery drawing for members who returned the postcards.

All participants were evaluated in annual physicals where they were weighed, their height was measured, their dietary intake evaluated using a standard 60-item Food Frequency Questionnaire, and they were asked about behaviors like exercising, eating fruits and vegetables, decreasing fat intake, using “unhealthy diet practices” like laxatives and diet pills or liquid diet supplements, weighing themselves, and smoking. At some point in the study, a questionnaire was administered to test “message recognition.”

Participation in the “intervention” group was high—68% of postcards were returned, 80% of the participants reported having read most or all of the newsletters at their annual visits, and 25% participated in one or more of the extra activities. The “message recognition” test was somewhat successful—the intervention group was significantly more likely to identify the 5 targeted treatment messages as being among the best ways to prevent weight gain; however, even 66% of the control group endorsed the treatment messages. The intervention groups were slightly-but-significantly more likely to weigh themselves and more likely to continue practicing “health weight loss practices” as measured by a 23-item questionnaire. However, changes in BMI, energy intake, percent of calories from fat, and rates of physical activity were not significantly different between the control and intervention groups. All participants gained an average of 3.5 lbs over the course of the 3 years.

In short, the intervention was a failure. The authors conclude:

It is easier to teach people what to do than to persuade them to actually do it…. The overall impact on weight itself…was very weak, indicating that stronger educational strategies are needed or, alternatively education alone is insufficient to deal effectively with this important problem.

Weight Maintenance, Behaviors and Barriers—Befort et al 2007 (free abstract or full text with umich login): Based on the abstract, this study sounds like a success, but under closer examination, not so much. The data was collected at a university weight loss clinic where participants were recruited to follow low-calorie or very low-calorie (500 kcal/day) weight-loss diets followed by a maintenance program. The “weight-loss” phase lasted for 3 months during which participants consumed prepackaged meals and/or shakes. The maintenance programs ranged from 6 to 21 months and consisted of weekly or bi-weekly meetings at the clinic during which participants were counseled to follow a structured diet plan with a daily calorie goal and exercise 150-300 minutes per week. In 3 out of 4 trials, the participants were also encouraged to continue consuming the shakes/prepackaged meals.

Out of 461 participants who started treatment, 44 dropped out during the 3-month weight loss phase and 211 dropped out during the maintenance phase. They sent follow-up surveys to everyone who completed the 3-month weight loss phase (n=417), and got 179 back (46.6.%). The more recently participants had been part of one of the studies, the more likely they were to respond to the follow-up survey. Responders had only been out of treatment for an average of 14 months.

Their claim that a “majority” of the participants maintained their initial weight loss is based on them lumping together respondents who had only been out of treatment for 6 months with people who had been out of treatment for 24 months or more, despite the fact that—just like in every other study of calorie-restriction weight loss—the results showed that most participants gradually regain weight. As they admit:

Compared to participants who were out from treatment for 24 months or longer, those who were out for less than 6 months (P<0.05) or for 6–12 months (P<0.01) had significantly greater weight loss maintenance, both in terms of kg and percent of baseline weight.

What they don’t say is that the percentage of respondents who report maintaining their initial weight loss drops off precipitously after 24 months.

no graph; perhaps it would have been too damning?

Of the 31 respondents who’d been out of treatment for 24+ months, only 25.8% had maintained a weight loss of 10% of their body weight or more and 48.4% had maintained a weight loss of 5% or more. That means out of the original pool of 417 who completed the 3-month diet, only 8 had proven capable of maintaining weight loss equal to 10% of their body weight for more than 2 years and only 15 had proven capable of maintaining a weight loss equal to 5% of their body weight. Other participants might be able to maintain their initial weight loss—that data isn’t available, but the trajectory certainly doesn’t look good. And that’s based on the half of the participants who participated in the follow-up—as Garner and Wooley note, the higher the rate of participation and the longer the follow up, the less weight loss on average is maintained.

What About the National Weight Loss Control Registry?

Several of the studies and at least one person who commented on one of the earlier posts in this series mentioned the National Weight Loss Control Registry (NWCR) as evidence that people can indeed lose weight and keep it off. I’ve never disputed that. Even in the studies that show the least hope for long-term maintenance, there are exceptions to the general trend. But that’s what they are: exceptions.

According to the NWCR website, they have over 5,000 members, all of whom have lost at least 30 lbs and kept it off for at least 1 year; however, most of them have done far better—registry members have lost an average of 66 lbs and kept it off for an average of 5.5 years. As the research above suggests, that’s not remotely “representative” of people who attempt to lose weight. On the contrary, the entire raison d’être of the registry is to figure out what’s different about the 5-10% of dieters who lose significant amounts of weight and keep it off. The goal is to identify strategies that might help other dieters, but as the researchers who run the registry admitted in a 2005 article (free abstract):

Because this is not a random sample of those who attempt weight loss, the results have limited generalizability to the entire population of overweight and obese individuals.

Indeed, the kinds of things the registry members do are generally the same things the participants in most weight loss studies are counseled to do (or, in clinical settings, forced to do): most of them follow a low calorie, low fat diet, eat breakfast every day, weigh themselves at least once a week, watch less than 10 hrs of TV per week, and engage in very high levels of activity—420 minutes per week on average. The NWCR has yet to figure out what makes those things work for them and/or makes them capable of sustaining those behaviors when for most people, they don’t/can’t.

Collecting 5,000 success stories does not prove that dieting “works” for most people let alone that it’s the norm. Somewhere between 45 million and 90 million Americans diet to lose weight every year, most of them by attempting to reduce their caloric intake. According to a survey conducted in April 2010 by a private consumer research firm on behalf of Nutrisystem, 30% of Americans have dieted repeatedly—an average of 20 times. Unsurprisingly, weight loss attempts are more common among overweight and obese people. If calorie-restriction dieting “worked,” America would be a nation of thin people.

Conclusion: Putting the burden of proof back where it belongs

Traditionally, researchers assume that a treatment is not effective until they have evidence that proves otherwise. The reverse is true in regard to weight-loss dieting: most people assume dieting is effective for long-term weight loss and challenge anyone who believes otherwise to prove that it doesn’t—not that that’s difficult, given the consistent failure of most weight-loss interventions to produce lasting results. I have not been able to find one long-term, randomized, controlled study that shows that dieting works (i.e. a statistically significant group of people following a calorie-reduction diet losing a clinically significant amount of weight and keeping it off for more than 3 years). Instead, what all the research to date shows is that the most reliable outcome of calorie-restriction dieting is short-term weight loss followed by weight regain.

I suspect the stubborn persistence in prescribing calorie-restriction dieting as a weight loss strategy in spite of the available evidence probably has a lot to do with dominant and deeply-engrained attitudes about fatness, meritocracy, virtue, and effort. People exhibit remarkable cognitive dissonance when it comes to the research on weight loss—they hold up exceptions as the rule and claim that the 90-95% of people for whom calorie restriction dieting does not produce weight loss must simply not be trying hard enough. 

Imagine this scenario playing out with any other condition—imagine that instead of weight, we were talking about some kind of rash that was widely considered unattractive and thought to be correlated with a variety of other health problems. There’s a treatment that showed promise in short-term trials. In virtually every study, most of the people who get the treatment experience significant improvement in their symptoms, with peak results around six months. However, in longer-term studies, there’s a reversal. Just as consistently, the vast majority of sufferers—at least 75% and usually closer to 90 or 95%—experience a gradual return of their symptoms. For approximately 30-40% of participants, their symptoms actually get worse than before they started the treatment. Only 5-10% show lasting improvement. Of course you would want to do more research to figure out why the treatment works for that 5-10%, but in the meantime, would you keep prescribing it to everyone with the same skin condition?

Even if the problem is that only 5-10% of them fail to use the treatment as instructed—say, it’s a topical cream that only works if you apply it every hour on the hour and people get fatigued, especially by trying to wake up at night to put it on. If 90% of the affected population can’t use the treatment effectively, the results are the same as if the treatment never worked in the first place. Well, except for that part where 30-40% of them end up worse off than before they started the treatment…

So even if the calorie counts on menus were accurate, and people could accurately and reliably estimate how many calories they burn, and they did choose lower-calorie options at least some of the time, and they didn’t compensate by eating more on other occasions…in other words, even if the calorie counts worked the way they were intended to, the best you could hope for would be short-term weight loss. There’s no reason to believe the policy—even under ideal conditions—would have a lasting effect on most Americans’ weight or health.

HFCS Follow-up: What the Rats at Princeton Can and Can’t Tell Us

Ed called my attention to last week’s press release about the study at Princeton currently getting some mass media attention. The press release claims:

Rats with access to high-fructose corn syrup gained significantly more weight than those with access to table sugar, even when their overall caloric intake was the same. 

i know it's a squirrel, not a rat. apparently no one's gotten a rat to do this and then circulated it with the right keywords to match my google search. this image likely not original to:’s pretty surprising, given that other studies have suggested that there is no difference between HFCS and sucrose. The Princeton study doesn’t offer a definitive explanation for the difference they found, but they suggest that it may have something to do with the slightly greater proportion of fructose in the HFCS.

As I noted in the first post on high-fructose corn syrup, HFCS-55, which is the kind used in soft drinks and the Princeton study, has roughly the same proportions of fructose and glucose as table sugar. Table sugar, or sucrose, is composed of fructose bonded to glucose so it’s a perfect 50-50 split. HFCS-55 contains 55% fructose, 42% glucose, and 3% larger sugar molecules. There’s a lot of evidence that fructose is metabolized differently than glucose, and may promote the accumulation of fat, especially in the liver and abdomen. Indeed, that’s why I believe that agave nectar is probably nutritionally worse than table sugar. Still, I’d be pretty shocked if a 5% increase in fructose could produce a statistically significant difference in weight gain, unless the rats were eating nothing but sugar-water. And they weren’t—in both of the experiments reported in the original study, the rats had access to unlimited “standard rat chow,”

Experiment 1: Rats Who Binge?

In the first experiment, 40 male rats were divided into four groups of ten. All of them had 24-hour access to rat chow and water. Group 1 was the control, so they just had chow and water. Group 2 had 24-access to an 8% solution of HFCS (.24 kcal/mL), which the press release claims is “half as concentrated as most sodas”. Group 3 had 12-hr access to the same HFCS solution. And Group 4 had 12-hr access to a 10% solution of sugar dissolved in water (.4 kcal/mL), which the press release claims is “the same as is found in some commercial soft drinks.” The two things of note so far are that none of the rats had 24-hr access to sucrose-sweetened water, and that the concentration of the sucrose was nearly 2x that of the HFCS syrup.*

Why the 24 hr vs 12 hr groups? According to the study:

We selected these schedules to allow comparison of intermittent and continuous access, as our previous publications show limited (12 h) access to sucrose precipitates binge-eating behavior (Avena et al., 2006).

In other words, they fed the sucrose group on a schedule that they already knew would cause binging. And they didn’t include a 24-hr sucrose group to control for that.

That helps to explain the results: the rats that had 24-hr access to HFCS-water gained less weight than either the rats who had 12-hr access to sucrose-water or the rats that had 12-hr access to HFCS-water. So according to the experiment, it’s better to consume some HFCS than it is to binge on sugar (not, obviously, how they chose to frame it in either the formal write-up or the press release).

Princeton rats

The only difference between the four groups in the first experiment that was statistically significant at a p<0.05 was between the rats who got chow only and the rats who got 12-hr HFCS. There was no statistically significant difference between the rats who had 12-hr access to sucrose-water and the rats who had 12-hr access to HFCS-water. There wasn’t even a significant difference between the rats who had 24-hr access to HFCS-water and the chow-only rats. So the only basis for the claim in the press release that HFCS is worse than sucrose is the fact that the rats with 12-hr HFCS got a “significant” amount fatter while the 12-hr sucrose rats didn’t. Even though the 24-hr HFCS rats didn’t either.

I am not the only one who’s picked up on this—both Marion Nestle (a vocal critic of the food industry) and Karen Kaplan (not, as far as I can tell, a shill for the Corn Refiners Association) also dispute the claim that this research demonstrates anything conclusive about HFCS vs. sucrose. The lead researcher replied to Nestle’s post, and rather than addressing the discrepancy between the 12-hr and 24-hr HFCS groups, he merely corrects her assumption that the 24-hr rats should be fatter:

There have been several studies showing that when rats are offered a palatable food on a limited basis, they consume as much or more of it than rats offered the same diet ad libitum, and in some cases this can produce an increase in body weight. So, it is incorrect to expect that just because the rats have a food available ad libitum, they should gain more weight than rats with food available on a limited basis. –Bart Hoebel

Which just makes it all the more baffling why they didn’t include a 24-hr sucrose group. Additionally, according to their results, binging or “consuming more” doesn’t explain the results, because:

There was no overall difference in total caloric intake (sugar plus chow) among the sucrose group and two HFCS groups. Further, no difference was found in HFCS intake and total overall caloric intake in the groups given 12-h access versus 24-h access. Both groups consumed the same amount of HFCS on average (21.3±2.0 kcal HFCS in 12-h versus 20.1±1.6 kcal HFCS in 24 h), even though only the 12-h group showed a significant difference in body weight when compared with the control groups.

The only explanation they offer for these results is the slight difference in the amount of fructose the rats in the HFCS and sucrose groups consumed. But even that relies on the idea that the HFCS rats did not feel as satisfied by their sugar water and compensated by eating more:

…fructose intake might not result in the degree of satiety that would normally ensue with a meal of glucose or sucrose, and this could contribute to increased body weight.

Unless satisfaction itself makes rats thinner.

Experiment 2 (Males): Wait, Where’s the Sucrose?

In the first part of the second experiment, 24 male rats were divided into three groups of eight. Again, all three had unlimited chow and water. Group 1 had 24-hr access to the HFCS-solution, Group 2 had 12-hr access to the HFCS-solution, and Group 3 was the chow-only control. Sucrose, you’ll note, drops out entirely. According to the study:

Since we did not see effects of sucrose on body weight in Experiment 1 with males, we did not include sucrose groups in this long-term analysis in males.

But there were no effects of HFCS on body weight on the 24-hr schedule! The omission of sucrose from this experiment makes as much sense as the omission of a 24-hr sucrose group in the first one. The lead researcher’s reply to Marion Nestle’s criticisms offered no further clarification about this choice. 

We explain in the article that we purposefully did not compare HFCS to sucrose in Experiment 2 in males, because we did not see an effect of sucrose on body weight in males in Experiment 1.

This study went on for 6 months instead of 2 months and, as the table above shows, the groups with both 24-hr and 12-hr access to HFCS-water gained a significantly greater amount of weight than the chow-only rats. This time, the 24-hr HFCS rats gained more weight than the 12-hr HFCS rats.

Experiment 2 (Females): Sucrose is back (still only 12-hr)! But chow is limited.

In order to “determine if the findings applied to both sexes,” they also ran a slightly different version of the second experiment on some female rats (n unknown). The control group, as usual, got unlimited chow and food. Group 1 got 24-hr access to HFCS-water. The remaining two groups got 12-hr access to chow (“to determine if limited access to chow, in the presence of HFCS or sucrose, could affect body weight”) and either 12-hr access to HFCS-water or 12-hr access to sucrose-water. Yeesh. How about testing one thing at a time, guys?**

So this time, only the rats with 24-hr access to HFCS gained a significantly greater amount of weight than the chow-only rats, which flies in the face of the claim that rats with limited access to a palatable food eat more. And the 12-hr sucrose rats actually gained slightly more weight (though not a statistically significant amount) than the 12-hr HFCS rats.

In other words, the findings in the three studies were completely inconsistent. For male rats in the short term, 12-hr access to HFCS induces significant weight gain but 24-hr access to HFCS does not. For male rats in the long term, both 12-hr or 24-hr access to HFCS induces significant weight gain, but they didn’t test sucrose. For female rats in the long term, only 24-hr access to HFCS with unlimited chow induces significant weight gain and limited chow, HFCS, and sucrose do not. And yet, based on this, they claim:

In Experiment 2 (long-term study, 6–7 months), HFCS caused an increase in body weight greater than that of sucrose in both male and female rats. This increase in body weight was accompanied by an increase in fat accrual and circulating levels of TG, shows that this increase in body weight is reflective of obesity.

Despite the fact that Experiment 2 didn’t even test the long-term effects of sucrose consumption on male rats, and 12-hr HFCS (albeit with limited chow) didn’t cause significant weight gain in female rats.

As Usual: Needs More Research

Based on the results of all three experiments, they conclude:

Rats maintained on a diet rich in HFCS for 6 or 7 months show abnormal weight gain, increased circulating TG and augmented fat deposition. All of these factors indicate obesity. Thus, over-consumption of HFCS could very well be a major factor in the
“obesity epidemic,” which correlates with the upsurge in the use of HFCS.

Despite the fact that obesity has also increased in many countries where HFCS is virtually never used, like Australia. According to a 2008 USDA paper:

Australia and the United States have a high and rising prevalence of obesity. They have opposite sugar policies: virtually no distortions affect Australia’s use of sugar, whereas sugar policy in the United States taxes sugar use. Sugar consumption per capita in Australia has been flat from 1980 to 2001, after which it increased by 10%-15%. Sugar is the major sweetener consumed in Australia.

The fact that the experiment doesn’t seem to show that HFCS is necessarily worse than sucrose doesn’t mean the findings aren’t intriguing. I really do want to know, for example, why rats with 12-hr access to HFCS gain more weight in the short term than rats with 24-hr access to HFCS, but the 24-hr HFCS rats gain more in the long term. And if, as they claim, the rats in all the groups consumed the same number of calories—which Nestle doubts because, "measuring the caloric intake of lab rats is notoriously difficult to do (they are messy)”—why were there any differences at all at the end of the trials? If none of the rats are eating more (and indeed, it seems that in some cases the HFCS rats were eating slightly less), what is the mechanism causing them to gain more weight, at least on some feeding schedules?

Does the concentration of the sugar have anything to do with it? In his reply to Nestle, Hoebel says:

Eating sucrose does not necessarily increase body weight in rats, although it has been shown to do so in some studies, usually employing high concentrations of sucrose, such as 32%. Our previously published work, has found no effect of 10% sucrose on mean body weight. At this concentration, rats seem to compensate for the sucrose calories by eating less chow.

I want to know if that’s true for HFCS as well. And did the difference in the concentrations of the HFCS and sucrose drinks have anything to do with the difference in the rats’ weight in this study?

Or does it maybe have something to do with sucrase, the enzyme that splits the fructose and glucose in table sugar? From what I’ve read, sucrase is present in the human digestive tract in sufficient amounts that it doesn’t rate-limit the absorption of those sugars in sucrose compared to the consumption of free fructose and glucose. But is it somehow involved in metabolism or appetite-regulation?

So rather than answering any questions about HFCS vs. table sugar, this really just raises a lot of new ones.

*It’s also not clear why they gave them different concentrations of sweetener. You’d think they would make them both soda-strength, or at least calorically equivalent.

**The failure to control for multiple variables does, in fact, complicate their ability to make any conclusions about gender difference:

In the present study, male rats maintained on 12-h access to HFCS also gained significantly more weight than chow-fed controls, while female rats maintained on 12-h access did not. It is possible that this can be accounted for by the fact that these males had ad libitum chow, while the females had 12-h access to chow. It is possible that the lack of chow for 12 h daily suppressed weight gain and TG levels
that might have otherwise been elevated in the female 12-h HFCS access group. This would indicate an effect of diet rather than a gender difference.

Salt Headlines That Make The Vein In My Forehead Throb

Salt has been all over the news this week because of a study just published in The New England Journal of Medicine claiming that if everyone in the U.S. reduced their sodium consumption by 3 grams/day, there would be 32,000 fewer strokes, 54,000 fewer heart attacks, and 44,000 fewer deaths every year. The story that got my attention was:

Remaining Arctic Ice Seen Melting Away Completely! (...on a computer screen)

That’s surprising, I thought. Everything I’ve read suggests that the relationship between salt consumption and cardiovascular disease is weak, inconsistent, and probably only valid for 20-30% of the population. So I expected the article to refer to some new research where, you know, “big benefits” were “seen.” As in observed. Like, in the world. And, given the claim about the magnitude, probably also measured.

To their credit, the authors of the study claim no such thing. The numbers are projections based on the application of several assumed effects of salt reduction, adjusted for different demographics and then applied to a model of the entire U.S. population. Thus, the title of the study: “Projected Effect of Dietary Salt Reductions on Future Cardiovascular Disease.”

The article seems to grasp the essentially speculative nature of the findings. The very first sentence uses the conditional tense:

…scientists writing in The New England Journal of Medicine conclude that lowering the amount of salt people eat by even a small amount could reduce cases of heart disease, stroke and heart attacks as much as reductions in smoking, obesity, and cholesterol levels.

The headline, on the other hand, seems to have confused the “scientists” with clairvoyants. Never mind doing any checking into the validity of their assumptions.

And the claim about how the benefits compare to smoking and obesity reduction led to a few headlines like this:

webmd salt

This crazypants idea initially sounds a lot like what the study’s lead author claims:

"The cardiovascular benefits of reduced salt intake are on par with the benefits of population-wide reductions in tobacco use, obesity, and cholesterol levels."

But the logic behind the claim is that a small improvement in the health of every single American would be as significant as a large health improvement in the health of every single smoker:

Dr. Bibbins-Domingo said that for many people the decrease in blood pressure would be modest, which is why, she said, “many physicians have thrown up their hands and said, ‘I’m not going to advise my patients to reduce salt because it’s too hard for patients and the benefits for any individual are small.’

“But small incremental changes in salt, such as lowering salt in tomato sauce or breads and cereals by a small amount, would achieve small changes in blood pressure that would have a measurable effect across the whole population,” she said. “That’s the reason why this intervention works better than just targeting smokers.”

For any given individual, there is no question about whether cutting salt is even close to “as good” as quitting smoking. The evidence for the link between smoking and lung cancer and death is strong, reliable, consistent, and has a clear causal mechanism (carcinogens). The link between salt and cardiovascular disease and death is weak, inconsistent, and still poorly understood.

That latter point starts to get at the problems with the study itself, and not just the headlines it inspired. A number of the assumptions the projection was based on are either demonstrably false or simply unsubstantiated. More on this some other time; for now, a few quotes and links to the essays they come from in Esquire and the medical journal Hypertension:

In a more recent statement, the founder of the American Society of Hypertension, Dr. John Laragh, goes further: "Is there any proven reason for us to grossly modify our salt intake or systematically avoid table salt? Generally speaking the answer is either a resounding no, or at that, at best, there is not any positive direct evidence to support such recommendations."

Studies show that 30 percent of the Americans who have high blood pressure would greatly benefit from a low-sodium diet. But that’s about 10 percent of the overall population — the rest of us are fine with sodium. And drastically cutting out sodium may actually hurt some people. ( "Go Ahead, Salt Your Food")


The available data suggest that the association of sodium intake to health outcomes reflected in morbidity and mortality rates is modest and inconsistent. Therefore, on the basis of the existing evidence, it seems highly unlikely that any single dietary sodium intake will be appropriate or desirable for each member of an entire population…. The decision to adopt a low sodium diet should be made with awareness that there is no evidence that this approach to blood pressure reduction is either safe, in terms of ultimate health impact, or that it is as effective in producing cardioprotection as has been proven for some drug therapies. (Salt, Blood Pressure, and Human Health)