Date: April 3rd, 2008

Speaker's Name, Affiliation:

Eva Kovacs
Unilever

Seminar Title: "Satiety Profiles of Foods: Assessment of Satiety Responses"

Presiding Chair: Harry R. Kissileff, Ph.D.

Rapporteur: Kathleen L. Keller, Ph.D.

Attendees and their Affiliation:

Kathleen Keller	Columbia/Obesity Research Center
Chris Ochner	Obesity Research Center
Allan Geliebter	Obesity Research Center
JA Grinker	University of Michigan
Rhoda Gruen	Columbia
Tony Sclafani	Brooklyn College
Gerry Smith	Weill Medical College
Isabelle Axelsson	ORC
Sheena Harris	ORC Columbia IHN
Doug Mook	N/A
Laurence Nolan	Wagner College
Steve Heymsfield	Merck
Tatiana Ungredda	ORC
Karen Ackroff	Brooklyn College
Michelle Iching Lee	ORC
Katherine Halmi	Cornell
Susan Carnell	ORC
Emmanuel Pothos	Tufts

Summary:

Many studies have shown that structured meal replacement plans are effective for short- and long-term weight loss and maintenance and for improvement of health risks associated with obesity. However, many dieters complain of hunger when dieting, which may lead to poor compliance and relapse. MR products that are designed to satisfy hunger can help consumers to more easily comply with a reduced-calorie diet.

For development and claim support of satiety-optimized products, behavioral testing in human subjects is required to provide a direct test of the benefits. Study design and statistical analysis provide the basis for objective claim support. Satiety studies typically compare the area under the curve, which cannot distinguish differences in the response profile or duration. The interpolation approach commonly used to estimate Time To Return To Baseline (TTRTB) only produces a single (group or treatment) mean value, since interpolation of individual curves is often impossible. Therefore, we aimed at defining a method that allows for quantitative analysis of duration of satiety response and allows for statistical comparisons among treatments. Data were derived from 8 studies with identical protocols assessing satiety responses of Slim?Fast meal replacements and other foods. On test days, subjects rated appetite-related parameters on line scales at baseline and at regular intervals for 300 min post-consumption of the test products (consumed as breakfast). Results were used as input for testing of various curve-modeling procedures. We have found that the Weibull function gives the best model fit and ability to determine mean TTRTB and 95% interval. Used in pharmacology, this function describes a biological model reflecting typical satiety responses.

We used this approach to evaluate the duration of satiety of a Slim?Fast ready-to-drink meal replacement shake (190 kcal) relative to other foods of equal (yogurt) or greater energy content (bagel meal or hamburger meal, 400 kcal). The TTRTB for hunger was found to be significantly longer (p<0.05) for the Slim?Fast shake (mean [interval], 306 [277-346] min) relative to yogurt (215 [205-228) min). When corrected for energy content, the TTRTB was significantly longer (p<0.05) for the Slim?Fast shake (1.61 [1.46-1.82] min/kcal] relative to yogurt (1.13 [1.08-1.20] min/kcal), bagel meal (0.81 [0.74-0.90] min/kcal) and hamburger meal (0.67 [0.61-0.74] min/kcal). The study demonstrated that the Slim?Fast meal replacement shake gives a level and duration of satiety that equals or exceeds that of other foods or equal or greater energy content. The potent satiety effect of the Slim?Fast meal replacement also suggests that caution should be used in making generalizations and assumptions about a poor satiety value of liquids vs. solids. It is likely that there are important differences between 'simple' beverages and more complex and nutrient-rich liquid foods, such as meal replacements, the latter possibly behaving more closely to solid foods with respect to satiety. We also used the Weibull approach as a basis for substantiating the "controls hunger up to 4 hours" claim for Slim?Fast meal replacements products.

It is concluded that the Weibull function gives an unbiased, quantitative basis for statistical comparison of duration of satiety responses. This can be used as a basis for substantiating appetite control claims based on duration of effect, provided that standardization procedures and transparent criteria are applied.

Discussion:

Q. Are the subjects using meal replacements during the entire period?
A. Yes, they get 1 meal replacement / day.

Q. Why didn't that group lose as much weight?
A. The first phase is different, so when they switched to 1 meal replacement per day, there is a different baseline body weight between the treatment and the control group.

Q. Which meal did they replace?
A. It's there choice, but typically breakfast or lunch.

Q. Do you know what your dropout rate is?
A. I'd have to check.

Q. Do you know how satisfied your subjects were with the meal replacements?
A. We didn't assess that.

Q. Do you have both subjects that want to lose weight, and those that don't?
A. Yes, but we try to eliminate people who have been on a diet in the past 3 months.

Q. How do you assess eating disorders and weight practices?
A. We use the TFEQ (Three Factor Eating Questionnaire) and the SCOFF questionnaire.

Q. How long is the line scale that you use to assess eating behaviors and reported sensations?
A. ~ 60 mm on the electronic device

Q. Does the device that you give to participants beep to remind them to make their ratings?
A. Yes, does. And, it's also time stamped, so when we are recording data, this is helpful. They can't make all of their ratings for the week on one day.

Q. Are all of your physical discomfort ratings assessed on the same scale?
A. For physical discomfort, we generally use a categorical scale (from 1 to 4).

Q. What are people doing between the time they eat meals and give their appetite reports?
A. We have things for them to do. They can either sit in the lab and read or work, or they go home. The main restriction is that they are not allowed to eat.

Q. Have you tried curve fitting your data?
A. I'll explain that in a bit.

Q. These are hunger curves, correct?
A. Yes.

Q. What do you do with the outliers?
A. What we try to do is first look at individual curves before de-blinding. Then, we go back to the individual subjects to see if there were some odd experimental conditions explaining their ratings. Based on that, we determine what must be outliers in the data.

Q. How many subjects do you typically have in your studies?
A. Around 24/study.

Q. What do you mean by "no physiological meaning for a minimum value (at time = 0 min)?
A. I'm maybe not explaining that so well, but I say that because the ratings do not use the entire scale. We are not really sure what those ratings mean.

Q. Is there any reason that you don't express each individual as a percent of their own baseline?
A. We've never tried that, but I'm not sure if it would make a huge difference.

Q. Most of the problems you are having deal with the issues of thresholds. Have you ever thought about calculating the half-lifes?
A. We haven't done that and we'd have to investigate that further.

Q. Why did you pick the yogurt?
A. We wanted to choose a product with an energy content similar to that of a SlimFast meal replacement, but something that you would not necessarily choose to consume as a meal.

Q. Do the time to return to baseline correlate depending on whether you look at desire for a meal or hunger?
A. Yes, hunger and desire for a meal correlate with one another.

Q. Do subjects in these experiments know what kind of drink they are getting?
A. Yes

Q. Do you think it's the protein in the drink? Or the fiber?
A. The reasons why the SlimFast meal replacement shows such a relatively high satiety value are not immediately obvious and may reflect a combination of features (e.g. protein content, fiber content, texture, viscosity). The product has been designed to optimize its satiety value through macronutrient composition and ingredient sources and use of proprietary technologies.

Q. What has changed about the drink formulation over the past 30 years?
A. Mainly the type and amount of sugar that is used, and the variety of SlimFast product available.

Q. Has this been replicated? It is baffling why yogurt would perform so poorly.
A. The meal replacement has fiber in it, so that is one potential reason why it is more satisfying than the yogurt. But, we have not yet replicated the findings. Comment: If you had a question on your scale that asked the subject how strongly they would like to eat a meal, you might be able to find the relationship between hunger and likelihood of eating.

Q. You aren't making any claims about what people will eat at the next meal, or long-term satiety, are you?
A. All this research is within the context of the SlimFast plan. Because meal replacements are meant to replace larger meals, they already provide an inherent form of control of energy intake. The goal is to enhance the satiety of the product to support compliance to the plan, not to reduce intake further.

Q. Do you have any data on SlimFast vs. meals that people would be replacing (comparing SlimFast to actual food)?
A. It is a difficult issue to define what a "typical meal" is, so we have not done these types of studies.

Q. Is it correct to use "satiety" in this context, or is this actually "satiation?"
A. Satiety is the correct term.

Q. Does it make a difference if subjects have used SlimFast before they come into your study?
A. We haven't looked at that.

Q. Are the panelists experienced in using the scales?
A. It's mixed, some are and some are not.