Transcript of Public Meeting:
Assessing Consumer Perceptions of Health Claims

Harvey W. Wiley Federal Building - Auditorium
College Park, Maryland
November 17, 2005

---

Proceedings

Welcome and Overview of Meeting

MS. FRASER: Good morning. If everyone could please take a seat? I am Leslye Fraser, the Director of Office of Regulations and Policy here, in the Center for Food Safety and Applied Nutrition, known to most of you as CFSAN. I have the privilege of serving as your moderator and time-keeper for today's public meeting.

Speaking for the entire Center, I would like to welcome you to this Consumer Studies Conference in which FDA and others will present the results of research aimed at assessing consumer perceptions of health claims. We believe we have planned a day that will be interesting and informative to all of us. CFSAN has invited several researchers from government, academia and industry to present their recent experimental findings, describing how consumers react to communications about substance-disease relationships.

Before beginning today, I would like to go over some housekeeping details. The restrooms are located right outside the door, at the top of the stairs and down the hall on the right as you walk toward the entrance where you came in. Food and drink is not allowed in the auditorium, and we invite you to participate or partake of any refreshments outside in the foyer. Lunch will be from 12:00 to 1:00 p.m. and may be obtained either in the Wiley Cafe, right outside our building, or in College Park on Route 1. There is also the Air Squadron Restaurant that has a buffet, right on Paint Branch Parkway several blocks around the corner.

We will have a 15-minute break in the morning and another 15-minute break in the afternoon. I ask you to return to your seats promptly as we have an ambitious agenda today. I also will ask you to turn off your cell phones or place them in silent mode if you have not done so already.

I now want to turn to the purpose of today's meeting, which is to discuss the findings from FDA's own and other research that examines consumers' reactions to health claims on conventional foods and dietary supplements. This is not a public meeting to discuss FDA's interim system for communicating the level of scientific support for health claims on conventional foods and dietary supplements, nor is it a meeting to address specific claims for which FDA may have determined it would or would not exercise enforcement discretion. Rather, this meeting is focused on the results of consumer research in this area and any implications of the available research for further consumer studies that may be needed or are already under way by other parties.

FDA intends to use the results of this research, as well as the comments we have received from stakeholders either here at this meeting or in writing, to determine what future actions and/or changes in policy may be appropriate. Toward that end, this morning and the first part of the afternoon primarily will be devoted to presentations of the various consumer studies that have been undertaken in this area.

To ensure everyone understands the background and underlying basis for the research that has been done, we will begin with three short presentations. The first will be from Louisa Nickerson, from FDA's Office of General Counsel, who will discuss the legal context of consumer research on health claims and other food labeling claims.

She will be followed by Dr. Barbara Schneeman, the Director of CFSAN's Office of Nutritional Product Labeling and Dietary Supplements, who will give a broad overview of FDA's scientific review of qualified health claims.

The third background presentation will be from Dr. Steven Bradbard, the team leader for CFSAN's consumer studies staff within my Office, who will help frame today's presentations by providing an overview of the various research methods used by the investigators you will hear today.

At that point, we will begin the presentation of the five consumer research studies. After each research presentation, and as time permits, we will allow a few minutes for clarifying questions. That is, if there is something in the presentation that you did not understand, you can ask questions of clarification and only those questions at that time. Following all presentations this afternoon, we have allotted two hours for attendees to make comments to a panel of FDA researchers that relate more broadly to the research that has been done or which you think needs to be done.

Dr. Brenda Derby, the principal investigator for FDA's qualified health claims study, will begin by presenting her research findings. Following that, we will hear from Dr. Pauline Ippolito, the Associate Director of the Bureau of Economics at the Federal Trade Commission; Wendy Reinhardt Kapsak, the Director of Health and Nutrition at the International Food Information Council in Washington, D.C.; Dr. Paula Bone and Dr. Karen France, from West Virginia University and, last but certainly not least, Dr. Neal Hooker from Ohio State University, regarding their respective consumer studies research.

When you came in you should have received a handout that provides a short biographical sketch of each speaker today and, in the interest of time, I will ask you to read it in lieu of my giving a more detailed introduction. I would, however, like to thank, on behalf of FDA, each of our visiting presenters for taking the time to participate in today's important public meeting.

Following a short break, we will allow attendees two hours to provide comments to our panel of FDA researchers. We ask that your comments focus primarily on recommendations for additional consumer research in the area of health claims. FDA is especially interested in hearing at that time your views regarding other schemes or signals that may effectively communicate to consumers the level of scientific support for health claims, without leading consumers to make erroneous inferences about the claimed substance-disease relationship and/or other product characteristics. We also are interested in hearing about alternative research methods that can empirically assess the effect or health impact on consumers' perceptions and behavior.

Again, I would like to reiterate that this meeting is focused on research that will assist policy makers, and during this portion of the meeting FDA panelists will be in a listening mode. Neither the panelists nor I are here to speak to the policies regarding health claims that may result from this research.

Lastly, I would like to note that over 20 have signed up to speak in advance in accordance with the procedures in the Federal Register notice announcing this meeting. Those persons will be given an opportunity to provide us comments first. You should have received a list of those speakers as you came in, which includes the order of presentation, and there are two names I need to add that, due to a glitch in our system, we did not get in, and I will give you those names later. We ask that each speaker limits your comments to five minutes maximum, and I will be monitoring the time to ensure that we can provide an opportunity for as many people as possible to comment. If you are still speaking when I signal to you that your time has elapsed, I ask that you cease speaking within 20 seconds to allow the next speaker a similar opportunity to provide comment to FDA. If there still is time after the speakers who requested an opportunity to speak in advance have made their presentations, we will offer others in attendance an opportunity to provide comments, subject to a time limitation that may be shorter depending on the requested number of speakers.

A transcript will be made of today's meeting's proceedings which will be available for public examination in our dockets. We also will accept written or electronic comments, including all relevant data and information related to the questions and the focus of this public meeting, at our dockets through January 17, 2006.

With that, it gives me great pleasure to introduce Louisa Nickerson, from our Office of General Counsel, to provide an overview of the legal context underlying food labeling claims. Again, welcome and thank you very much for joining us today.

Legal Context of Consumer Research on Health Claims and Other Food

[see presentation slides for Ms. Nickerson]

MS. NICKERSON: Good morning. As Leslye Fraser said, I am here to provide some legal context for the consumer research that you are going to hear about today. I would like to emphasize that the purpose of this meeting is to discuss the consumer research that FDA and other organizations have been doing and that the information I am about to provide about legal context is for background only. We are not going to be discussing the legal issues surrounding food labeling claims today. There will be other opportunities for you to comment on how FDA regulates health claims and other food labeling claims.

Well, let's start with what a health claim is. In FDA parlance, a health claim isn't what one might assume from the ordinary meaning of those words, in other words, any claim about health. We define it more narrowly as an express or implied statement in food labeling about the relationship of a food substance to a disease or health-related condition. Health claims can be made in the labeling of dietary supplements as well as conventional foods, and they require FDA review before they can appear on products in the marketplace. That usually takes place through a petition process.

The key elements of a health claim are, first, a substance, which is a specific food or component of food, whether that food is conventional food or in a dietary supplement form. The second element is a disease or health-related condition, which means damage to an organ, part, structure, or system of the body such that it does not function properly, or a state of health leading to such dysfunctioning, except that nutrient deficiency diseases are not included in this definition.

Some examples of substance--tuna, which is a specific food and omega-3 fatty acids, which are food components; an example of a disease--colon cancer; an example of a health-related condition, a state of health leading to disease, would be adenomatous colon polyps.

The purpose of health claims was to allow foods, including dietary supplements, to bear certain science-backed claims about disease prevention in their labeling without being regulated as drugs. To that end, in 1990 Congress passed the Nutrition Labeling and Education Act which gave FDA specific authority to permit health claims in the labeling of foods. I say certain scientific science-backed claims about disease prevention because health claims do not encompass claims of absolute prevention. In other words, "take Ultimate tablets and you'll never get cancer." They are simply risk reduction claims.

The reason why, before the NLEA, foods that bore claims even about reducing risk of a disease ran the risk of being regulated as drugs is that the drug definition includes, among other things, articles intended for use in the diagnosis, cure, mitigation, treatment or prevention of disease in man.

To give some examples, a claim that a product "prevents colds" would be a prevention claim and would make a product a drug. "Effective arthritis pain relief" would be a mitigation claim. "Wipes out gout" is a disease cure claim. "Shrinks tumors" is a disease treatment claim. Contrast these types of claims with health claims, which are always phrased in terms of "may reduce the risk of" some disease or health-related condition.

There was a federal appellate case a couple of years ago in which we litigated the scope of health claims, and the court upheld our interpretation that health claims are about reducing the risk of a disease or health-related condition and not about treating, mitigating or curing diseases.

The research being discussed today involves two different types of health claims, so-called unqualified health claims and qualified health claims.

Unqualified health claims are also referred to as SSA health claims, where SSA stands for significant scientific agreement, and that comes from the Nutrition Labeling and Education Act's standard for FDA to authorize health claims by regulation. It is significant scientific agreement among qualified experts. Sometimes these health claims are referred to as NLEA health claims. FDA authorizes them by regulation. Unqualified is a bit of a misnomer because they actually do contain the word "may" which can be viewed as a qualifier but, actually, that word is intended to indicate that the benefit is risk reduction population-wide as opposed to something that we know will occur in every single person.

Some examples of unqualified health claims that I have put up on the screen are authorized health claims for oat bran and heart disease, and low sodium foods and high blood pressure.

Qualified health claims are health claims that are based on scientific evidence that is credible but that does not meet the significant scientific agreement standard. These health claims include a disclaimer or other qualifying language to prevent consumers from being misled about the level of support for the claim or other important facts, which could be, for example, conditions of use that are necessary to get the risk reduction benefit. Qualified health claims are considered under FDA's exercise of enforcement discretion, and we do not authorize them by regulation.

Here are some examples of qualified health claims. One is for omega-3 fatty acids and heart disease. The other is for calcium supplements and hypertension or high blood pressure. Qualified health claims came about as a result of several court challenges under the First Amendment to FDA's denial of certain health claims that Congress directed us to consider in the NLEA. The original case that led to qualified health claims is called Pearson versus Shalala, and it was decided by the D.C. Circuit in 1999.

Today's research is going to involve some other types of labeling claims, specifically, structure/function claims which are claims about how a substance or product affects the structure or function of the human body, an example being that calcium helps build strong bones. Then, dietary guidance statements which are recommendations about dietary patterns and practices that promote health. So, they are more general than health claims are. An example of a dietary guidance statement is "eat plenty of fruits and vegetables every day for good health."

So, why are we interested in consumer research about food labeling claims? Well, first of all, we want to understand how claims in food labeling affect consumer perceptions and, more specifically, what kinds of claims in food labeling have the potential to mislead. Then, as a result of Pearson, we also need to determine whether potentially misleading health claims can be cured by disclaimers or other qualifying language. And, if the answer is yes, what are the characteristics of an effective disclaimer?

That is my presentation for this morning. Thank you very much.

[Applause]

MS. FRASER: Thank you, Louisa. We will next hear from Dr. Barbara Schneeman, the Director of the Office of Nutritional Product Labeling and Dietary Supplements, also known here as ONPLDS, and she will talk to us about scientific review of the qualified health claims that Louisa just talked about.

Overview of Qualified Health Claims (QHC)

[see presentation slides for Dr. Schneeman]

DR. SCHNEEMAN: Thank you, Leslye. Welcome to CFSAN, everyone, who is participating in this meeting. As Leslye has indicated, it is my task to give you a very brief overview of the steps that FDA has taken to develop a framework for the scientific review of qualified health claims.

Just to start first of all with putting some of Louisa's comments into context, certainly we are interested in health and preventing the incidence of disease or lowering the risk of disease. Qualified health claims then address those steps of how do we intervene to decrease risk of disease and, once something is in the box of treatment of disease, then that is where the drug category comes into play. Certainly, we recognize that either looking at what factors might increase risk or, conversely, to decrease risk we can use biomarkers, and there are several validated biomarkers that are available for assessing how certain dietary factors may reduce risk of certain types of disease.

So, just again to reiterate some of the background, in 1990 the Nutrition Labeling and Education Act was enacted, and this allowed for health claims on foods that were based on significant scientific agreement. In 1994 the Dietary Supplement Health and Education Act was enacted. This then provided for different types of claims as well, structure/function claims, claims of general well being and nutrient deficiency claims. Then, in 1999 we had the beginning of the Pearson court cases which did recognize the First Amendment protection of commercial speech. It provided for claims that did not meet the standard of significant scientific agreement if they were properly qualified to not mislead consumers. There were several claims that were allowed through the court decision.

So, in October of 2000 FDA began the process of developing its framework for scientific review of qualified health claims. The first step was to revoke the regulation that did not authorize four claims in the original NLEA regulations. As indicated, through the court decisions qualified health claims were subsequently allowed through enforcement discretion for antioxidant vitamins, the 0.8 mg of folic acid, and certain B vitamins related to vascular disease.

After the court decision, but before the implementation of our interim guidelines, there were several claims reviewed and enforcement discretion exercised. The omega-3 fatty acids in heart disease for use of dietary supplements; selenium in cancer; nuts in heart disease; walnuts in heart disease; and phosphatidylserine related to cognitive dysfunction and dementia.

Then, under former Commissioner Mark McClellan, a task force was convened, with the title of Consumer Health Information for Better Nutrition. This task force released its report in July of 2003, and in that task force report interim procedures were established for the review and evaluation of qualified health claims on both conventional foods and dietary supplements. Contained within that interim guidance is an interim evidence-based ranking system for evaluating scientific data.

At that time, there was also a consumer studies research agenda that was proposed and, clearly, one of the objectives of this meeting is to focus on the results of that consumer studies research agenda. CFSAN continues to have as a high priority developing a regulatory strategy for qualified health claims, but in the interim we are going through the process of reviewing health claims.

So, we then have the two circumstances, first of all, the claims that are authorized under NLEA that require rule-making by the agency, those that are based on significant scientific agreement, and the agency is still working on those claims. We actually have two under review right now, one about beta-glycan and barley and one about vitamin D and calcium related to osteoporosis. But now we also have qualified health claims, claims that characterize the quality and strength of the scientific evidence if the claim is not based on significant scientific agreement, and those are done through enforcement discretion under the interim guidelines.

Part of our way of thinking about the scientific review process is to think about the continuum of scientific evidence in developing our framework for qualified health claims. It is not that we are looking at different types of scientific evidence for qualified health claims versus SSA claims, but it is more the continuum of scientific evidence and the strength of that evidence in support of the claim. Obviously, at the highest level one could achieve scientific consensus. Significant scientific agreement is not quite at that high a level but certainly is a high standard and then the world of emerging evidence would allow us to make qualified health claims if suitably qualified.

Some of the language that we have in terms of qualifying language to characterize the level of scientific evidence--in our interim guidance we note that a significant scientific agreement claim would be equivalent to an A-level claim. Then we use the terminology of B-level claims, C-level and D to indicate decreasing levels of comfort relative to the science supporting the claim.

The slide here just gives you examples of the types of qualifying language that we had proposed in the interim guidance. However, it was noted there that the precise language may vary depending on the specific circumstances.

In that interim guidance document there was a set of steps outlined that the agency goes through in terms of defining the substance and disease relationship consistent with the regulations that Louisa has discussed. We identify the relevant studies. We want to look at the totality of evidence that is available. We classify those studies in terms of the nature of the publication itself--is it a review article? Is it a clinical study? Is it animal data? In vitro data? We rate the studies for quality. We look at factors such as what type of information it gives us in terms of dietary assessment, the statistical analysis used, are appropriate controls used so that we can, in fact, draw reasonable scientific conclusions from the body of evidence.

So, we rate those studies for their quality and then, once we have what we would define as usable information, now we want to rate the strength of that body of evidence relative to the quantity, the quality, consistency and relevance so we can come up with a rank in terms of the nature of the scientific evidence supporting the claim.

Actually, this can now be updated and I am not sure which you have as the handout, whether you have the updated slide or this one, but under the initiative, the consumer health initiative, we have received 17 petitions for qualified health claims. Many of these petitions contain multiple claims for review. We have had up to 30 claims in one petition, and each one of those claims has to be reviewed for the scientific evidence individually. So, at this point the total number of proposed claims that we have reviewed is actually 75 because we just issued a letter recently. Of those, FDA has considered exercising its enforcement discretion for, I believe it is 16 claims at this point.

In terms of the decisions, they are based on the review of the scientific evidence, and the wording of the claim then reflects the nature of the evidence that has been identified. In some cases our wording may actually describe conflicting data if that is the nature of the evidence that we have identified. Where we have not exercised our enforcement discretion, it is because we have been unable to identify credible scientific evidence in support of the proposed claim and so that was the decision that we came to.

So, in terms of getting more detailed information, we recommend that one looks at the interim guidance which has been on the web site since that report was published. But we also encourage you to look at the letters that describe our exercise of enforcement discretion. There is a section in each letter that lays out the review of scientific evidence that the agency goes through and gives you our current thinking in terms of the evaluation of that body of scientific evidence. So, it is a great resource in terms of understanding the process. With that, I will turn it back to Leslye. Thank you.

[Applause]

MS. FRASER: Thank you, Barbara. Between Louisa's and Barbara's presentation I think you have at least a background from a legal perspective as well as a scientific perspective of how FDA is evaluating health claims for conventional foods and dietary supplements. We are very thankful that Barbara was able to join us. She will be leaving shortly to catch a flight for an international meeting. So, if you see her duck out, that is where she is headed. Our next person, and this is the last piece of background information, is Dr. Steven Bradbard. He is within CFSAN's Office of Regulations and Policy, Division of Social Sciences, and he heads up the team of excellent consumer researchers here, in the Center. So, I will ask Steven to come forward and just give an overview of the types of research generally that people in this field have been engaged in.

Overview of Today's Consumer Studies Research

[see presentation slides for Dr. Bradbard]

DR. BRADBARD: Good morning. As Leslye mentioned, my name is Steve Bradbard. I am the team leader for the consumer studies staff here, in CFSAN, and I would like to extend a special welcome to today's presenters and to our invited guests on behalf of my team who are a very talented, multi-disciplinary group of professionals who have close to 200 combined years of social science experience. Alan Levy has 180 of those.

[Laughter]

Over the next few hours you will be hearing several investigators report findings from their recent research on qualified health claims. We realize that for many of you research methodology and data analyses are not part of your everyday work. With that in mind, I want to provide you with an overview of experimental research methods, which I hope will give you a framework for better understanding these study results that you will be hearing today.

I will briefly talk about some commonalities shared by these different studies; then talk about the topic of independent and dependent variables; touch on how researchers interpret data from experiments; and then point out to you the other types of factors that can influence study results.

All of these studies are experiments. They are designed to measure cause and effect relationships. What I mean by that is consumers are shown a health claim on a food label or ad copy for a product that states a substance-disease relationship. How does that information affect or influence their reactions when they are asked questions about that information? For example, consumers in all of these studies were shown a statement about a substance-disease relationship and then given a question such as, based on this, what is the level of scientific support for the claimed substance-disease relationship? We can measure, based on their ratings of the levels of scientific support, how they react to that statement. So, that is the cause and effect. The manipulation is the information about a substance-disease relationship; the effect is the rating on a measure based on a question that we ask.

Now, again, all of these studies are experiments and there are other research methods that you've probably heard about before, including surveys and focus groups. I have been to meetings where I have heard people using these different terms interchangeably as though they are all the same thing. In fact, it is very important that you understand that these are very different research methods and that an investigator will select a method based on the objective of a study.

None of these methods in itself is better than another method. The question is which method is the most appropriate to study the question that you are asking. Focus groups--and many of you have possibly participated in or observed a focus group--focus groups provide an in-depth qualitative understanding about the attitudes, beliefs, feelings and motivations that consumers may hold for a particular topic. For example, focus groups would be appropriate to use in order to evaluate consumers' immediate gut level reaction, their thoughts and feelings in response to terms such as 'level of scientific support.' So, in a focus group you might write on a flip chart "what do you immediately think of when you see level of scientific support?" And, you will get the participants pouring their hearts out about level of scientific support--hopefully. If not, you will prompt them to.

Surveys, in contrast, are used to provide population estimates of awareness and knowledge about a particular topic. For example, if an investigator was interested in estimating the percentage of Americans who have heard of qualified health claims, then a survey would be a useful tool.

Experiments, as I mentioned earlier, are designed to demonstrate cause and effect. So, if a researcher wants to evaluate how consumers react to different ways of qualifying a substance-disease relationship, then an experiment is the most appropriate method. So, no one will leave here today saying they heard interesting things about survey research. Right? Good.

During the presentations today you will likely hear the term independent variables. These refer to the label or ad characteristics that are manipulated in the experiment and then tested. Researchers can vary the way in which the level of scientific support is presented for a claimed substance-disease relationship. So, by example, they can test how consumers respond when words are used to qualify the level of scientific support. They could also test how consumers react to letter grades as qualifiers and they can evaluate what happens when you use a combination of words and letter grades.

Now, the level of scientific support is just one independent variable that can be studied in this type of experiment. It is also possible to vary the type of statement being tested. For example, how do consumers respond to health claims as opposed to structure/function claims as opposed to dietary guidance statements? The number of claims present on a label can also be manipulated, and you will hear in Dr. Hooker's study that he actually had two claims on his label as opposed to a single claim on the label.

Other variables such as the amount of information provided to consumers before they are asked to make judgments about a claim can be varied. In Dr. Derby's study you are going to hear about a full information control condition. This involved providing subjects with information ahead of time about the scientific support for the substance-disease relationship, or the lack of support for that relationship, and then evaluating how these newly educated consumers reacted to the claimed statements.

You will also hear the term dependent variables. Sometimes these are also referred to as outcome measures. In a cause-effect relationship the dependent variable is the measured effect. There is a variety of possible outcomes that can be measured, much like there is a variety of independent variables that can be manipulated. You will notice today that not every study uses the same dependent variable. Again, there is no single correct variable; it has to do with what it is that that investigator is looking to get information about.

The information about substance-disease relationships that is communicated by a label claim or ad copy can affect consumers' judgment and inferences about the level of scientific support for a health claim. All of these studies have some outcome measure related to level of scientific support. But you can also get information from consumers about their perceptions on the likelihood of actually obtaining the claimed health benefit; the overall healthfulness of the product itself; the quality and safety of the product; and their reported intent to purchase the product. All of these, again, are outcome measures.

So, to tie this together, these studies measure consumers' reactions to how the substance-disease relationship is communicated or expressed on a product label or in ad copy. This is accomplished by first showing them information about the substance-disease relationship, the independent variable, and then asking them to provide a rating in response to a specific question. Here is an example, "on a scale of 1 to 5, where 1 means very uncertain and 5 means very certain, how certain are scientists that substance X may reduce the risk of disease Y?"

Typically, in this type of research the participants will see one or maybe two different claim statements and then answer questions that require that they make judgments or inferences based on the information in those statements. When researchers analyze the ratings that the participants provide on the outcome measures, they make comparisons between the average scores for different groups of subjects. They don't look at any person's individual score. In doing the statistical analysis you compare the average scores for different groups.

For example, a study could have 60 consumers in group 1 see a substance-disease relationship with a strong disclaimer. Just by example, lets say a D-level disclaimer, which is intended to convey that the level of scientific support for that claimed benefit is weak. Sixty different consumers in group 2 can see the same substance-disease statement but with a weaker disclaimer, let's say a B-level disclaimer, which is intended to convey stronger support for the claimed benefit. All the participants provide individual ratings. We are making group comparisons but they don't all sit together in a room and come up with a single rating for 60 people. They provide their individual ratings and then we take the mean of those ratings. When researchers evaluate the data, they use statistics to determine if the average rating for group 1 on an outcome measure differs significantly from the average rating of group 2 on that same measure and that is how analysis is done.

Of course, there are other possible variables in an experiment, aside from the independent variable and the manipulated variable, that can have an effect on the outcome measure. For example, a person's gender, age, race, educational level, income status, even their geography because people living in certain parts of the country, say, may have different types of reactions to a particular type of message than people in other parts of the country--all of these things can possibly influence subjects' responses.

You will also hear today from Drs. France and Bone about other factors, such as claim-specific knowledge and attitudes; history of product use; the product-specific benefits; attitudes toward government and industry; and a person's health status. All of these things can affect consumers' judgments and inferences and must be taken into account during the analysis.

This is I think an important reminder because, you know, I have possibly bored you to death saying over and over again substance-disease relationship, substance-disease relationship. We feel strongly--researchers feel strongly that these are communication studies. This is communication research that uses experimental methods to evaluate how consumers react to different ways to discuss substance-disease relationships. That is what this research is about. The focus is not on the actual substance-disease relationship being used, but on the different ways in which this relationship could be presented. Thank you.

[Applause]

MS. FRASER: Thank you, Steve. We are doing very well on time so far, which is good--not to put pressure on you, Brenda. A couple of things, just housekeeping again or corrections, Barbara mentioned a couple of slides with updated numbers. We will have corrected slides if we can get it all worked out and copied for you either right before lunch or right after lunch.

I mentioned earlier that there were two speakers who had requested an opportunity to speak ahead of time, consistent with the Federal Register, that we inadvertently left off the list. With my apologies, please add as number 20 Dr. Iona Carabin, C-a-r-a-b-i-n, I-o-n-a, M.D. She is President and Medical Director of the Women's Health Sciences Institute. Number 21 will be the Burdock Group and Dr. Berna, B-e-r-n-a, Magnuson would be the speaker, M-a-g-n-u-s-o-n, Ph.D., doctor and toxicologist. So, those will round out the 21 speakers who we will ask to speak first.

With that, we will now get to the heart of what we are here to discuss, and Brenda Derby, within the Office of Regulations and Policy, under Steve's leadership here, will be discussing the results of FDA's consumer research findings. Brenda?

Presentation of FDA Results

[see presentation slides for Dr. Derby]

DR. DERBY: Good morning. Now we are going to get to the meat of what everybody came to hear today, and I do have a lot of slides at the beginning where I give an overview to show you the questions that we asked, the disclaimer schemes that we tested, and so forth, and I will try to go through things somewhat quickly so that we can get to the results. But I will also alert you that the detailed report is available on the web, and the last page of your handout shows you the address so, where we don't have charts, you can find those charts on your own at the web site.

As Steve mentioned and we like to emphasize, we have to use health claims, substance-disease relationships, in these studies but that is not what we are interested in. We just picked ones that seemed reasonable to use in our testing but they are not the focus, and what we are really interested in is assessing the effectiveness of different ways you can go about expressing any substance-disease relationship to indicate the level of scientific support for that claim.

When in shorthand I say health claim, all I am referring to is that simple statement that X may reduce the risk of Y, not all the other things that go with health claims within the regulations and within the law, and unqualified health claim condition means just that, that simple statement. When we have a qualified claim it is that statement plus one of the different ways of qualifying it that I will be showing you in just a moment. We have four disclaimer schemes that we tested in this study. Of course, there is an infinite number that people could think of. We tried to come up with some that seemed reasonable from what people are doing now and have suggested.

The four substance-disease relationships that we used are shown here, and they are in rank order, A, B, C, D. Calcium may reduce the risk of osteoporosis is a claim that has been through review and is an SSA type of claim.

The other three, although some of them have been through review since we designed the study, at that time had not. So, we just picked levels that seemed kind of reasonable. So, omega-3 fatty acids in the risk of heart disease is our B level claim. Selenium in cancer is a C level, and lycopene and cancer is D level. We looked at four possible disclaimer schemes and they were intended to convey the B, C and D levels of qualified claims, where B would be greater scientific support than C and than D.

The four schemes which have some kind of curious names, but I will show you examples and it will make more sense: point-counterpoint, which is using the words themselves to convey scientific support and where the claim is stated first, followed by the disclaimer statement. The embedded way of expressing it is to start with the disclaimer and then give the statement of the relationship. Then two report card schemes, one that does it just with text and one that does it with a graphic, and I will show you examples of each one.

Point-counterpoint would be something like "calcium may reduce the risk of osteoporosis," the claim first; "the scientific evidence is promising but not conclusive." In our study that would be a B level disclaimer.

The embedded scheme is a C level, "limited and inconclusive scientific evidence suggests that omega-3 fatty acids may reduce the risk of heart disease." So, here you are getting the disclaimer first.

The text report card tells you about the system: "a diet high in selenium may reduce the risk of cancer. FDA evaluated the scientific evidence and gave it a C rating, based on a scale from A, strongest evidence, to D, weakest evidence."

I should mention that in our experiments we never had an A anywhere. We show it on the graph that you are going to see here but we did not have a condition in which the A was ever checked because, under our current scheme, an SSA claim does not have anything indicating that it is that type of claim.

As Steve mentioned, this is an experimental design. We did it in shopping malls, five malls across the country and recruited 1,920 adults, an equal number at each site. These are volunteers, and each respondent is assigned to a condition. In our case, they saw two products but they never saw two disclaimers. One of the products they saw would be a control condition, and I will be showing you what those control conditions were in a moment. One of the conditions would have had a disclaimer. And, we counterbalanced the order.

We had to select foods to go with the four claims we selected and we wanted foods that are pretty commonplace and most people would have experience with. So, for the calcium osteoporosis we did a fortified orange juice. For omega-3 we did tuna but we didn't do the tuna can, we did the foil packs so that you would be able to read the claim; eggs for selenium and spaghetti sauce for lycopene. I will show you pictures. In your handout it is going to be pretty tiny. But one thing I should mention is that the only thing we varied are the claims. The nutrition facts panel is identical across all conditions and all other information is identical.

This one for the orange juice shows a control label where there are no claims at all. So, it just tells you the name of the orange juice. It says "fresh-squeezed" and at the bottom, where you probably can't read it, it has "100 percent pure Florida squeezed orange juice, not from concentrate," and then "64 oz." So, that part would be on every label for orange juice.

For the tuna, this one does have a claim on it. It is a B level claim in this example, and for each condition this is where the claim would be. Except for the full control, they would also have a nutrient content or declaration for the nutrient that is being mentioned in the claim.

Here is the eggs, and this is the selenium. It would always be in the same spot for each of the different conditions.

And the spaghetti sauce. That is the graphic that you saw earlier where the D is checked. But that is not the only one they see for the spaghetti sauce. They do see a C level claim as well.

We have four key measures of the impact of information on consumers. We call them performance measures. Steve mentioned some of these, the perceived strength of science; the perceived likelihood that you would get the claimed benefit if you ate this food; the perceived likelihood you might get other benefits not specifically mentioned on the claim; and then how important this food would be as part of your total diet. That is the more global rating of people's response to this information.

I will go through each one individually. For the strength of science we have a seven-point scale, from very uncertain to very certain. They were asked how certain is the scientific evidence that eating foods that contain whichever nutrient it was will reduce the risk of--and then the health condition.

For the claimed benefits we have one multi-part question that was used for both those measures. For the claim relevance it would be whichever one of these is relevant to the health claim, and then the others would be the ones that are other health benefits. So, they are asked this for each product they saw: reduce the risk of having a heart attack, high blood pressure or getting cancer and getting osteoporosis, again, on a seven-point scale, from not at all likely to very likely.

The final measure, the importance as part of your diet is a seven-point scale again, not at all important to very important. The question was worded, how important would this food be as part of a healthy diet for you, emphasizing the total diet aspect of eating this food.

I mentioned we had control conditions. We had quite a few control conditions. We had the no claims condition that you saw on the orange juice label a moment ago. We have the unqualified health claim statement, X may reduce the risk of Y being the only thing on there. Or, a condition where there was only a content claim or a nutrient declaration but no health claim. All the health claim conditions did include the nutrient claims.

We heard earlier that "may" sometimes has a connotation of being qualified. We have heard that in focus groups with consumers. So, we decided to look at that in this study so we reworded the unqualified health claim statement without the word "may." So, instead of saying "calcium may reduce the risk of osteoporosis," we said "calcium reduces the risk of osteoporosis," still indicating that it is a risk reduction type of claim but getting rid of the "may."

Then we had our full information condition. A quarter of the samples were randomly assigned to one of the four claims, and before they saw any labels they were asked to read one page. I think the copies are either in your packet or they are certainly upstairs, outside. We tried to write these in such a way that the A, B, C, D claims would convey different levels of scientific support, with the D the most negative, indicating the most weakness in what support there was. They didn't see a disclaimer; they saw the label with just the content declaration, and that nutrient was specifically mentioned in the full information so that they would know that there was a link between what they read and this product.

We identified for our analysis three performance standards to look at the effectiveness of these disclaimer schemes. The first was an obvious one, is there a linear effect? We expect that if it is a B claim it should get a higher rating of scientific support than a C claim or a D claim. So, we were looking to see whether that happens.

Then, what is the effect of a disclaimer of any kind on your perception of scientific certainty? As you would expect, there should be some relationship there.

Then, do these disclaimers have any effect on how people perceive the product? Because one of the things that is important about food labels as opposed to other places people get information about diet-disease relationships is that it is on a particular product so it is implying something about that product. It is not just about the science. I think we found in earlier research that that comes into play when people are deciding what they think of these claims--does it seem right for it to be on that product? Does this product have the right nutrient? Does that aspect of it seem credible?

So, our questions are do disclaimers convey decreasing levels of scientific support? And, we look for that linear effect. Does the scientific certainty rating change if you have a disclaimer versus not having a disclaimer? We expect it would counteract the effect of an unqualified statement when people do their ratings. And a comparison we do in the analysis in the reporting is we also compare what is the effect of having a claim versus not having a claim. So, first, do claims make a difference? Then, if claims make a difference, does adding that disclaimer to the claim change people's perceptions? So, we look at the inferences about other benefits than the questions that I mentioned earlier.

Finally the results--this is the only chart you are going to see today. I noticed that in the handout, for some mysterious reason, you don't have the labels on the axes but we have level B, C and D level disclaimers. On this side it is a normalized score of the scientific certainty rating. The nature of our design required us to normalize this, not a scale from 1-7. It is a zero mean and plus/minus one standard deviation.

This green is the text report card and this blue is the graphic report card. Those do what you would expect a disclaimer scheme to do. B is the highest, C is in the middle and D is at the bottom, and both of those relationships are significant at the 0.02 level or less.

The other two, the point-counterpoint and the embedded that relied on words to convey whether the scientific support was less, are not significantly different and don't have that nice slope that you would want. So, in other words, whatever they read, they gave it similar ratings. They didn't perceive less scientific certainty with a D level claim in words than they did with a B level claim.

So, those didn't work and, as a result, in all the subsequent results I am going to tell you about we dropped the people who were in those two conditions. Since the disclaimer was not conveying different levels of science there is no point having all that noise in your analysis. So, the subsequent analyses are all done just on those people who saw a report card in either a text form or a graphic form.

Because I am trying to cover a lot of information in this fairly short amount of time, you are only going to see bullets for the rest of them because, as I said, the charts are all in our report online.

I mentioned the two main effects that we are going to be looking at across our measures, whether it is an effect of the health claim itself compared to no claim and then the disclaimer versus an unqualified claim.

When we look at scientific certainty there is a health claim effect. People think the science is more certain when they see an unqualified health claim versus if you ask them about a product that just has a nutrient content claim. This effect is strongest for those nutrients they knew the least about. Earlier, in the recruiting part of the study, we asked people if they were familiar with the four nutrients that are included in our claims, and we found that, as you would expect, virtually everybody has heard about calcium having health effects. Omega-3 was about half. Then you get down to about a third when you are talking about selenium and lycopene, just having that basic recognition that this may have health effects. So, those less familiar nutrients having an unqualified claim had more of an impact on the reading of the science.

For disclaimers, the presence of the appropriate disclaimer was only effective for the lowest level claim, the D level lycopene claim. Surprisingly, level B and C claims were more positive with a disclaimer than without. So, that was kind of an unexpected finding.

When we look at what they think about the relevant health benefits, the claim is promising. Again, you find people were more positive when they saw a health claim, and more so if it was something less about a nutrient they were less familiar with. But for disclaimers we got no significant differences. So, people rated the relevant health benefit that the claim talks about the same whether or not it was a qualified claim. The disclaimer being there made no difference.

We looked at the potential for some halo effects, people in the past have found that the presence of a claim on a food product sometimes makes people think it has other good attributes that aren't mentioned. They think it just must be a good food and so they tend to give higher ratings to other questions. We only found that on half of them, two of the four, and those were--let me check, I think it was the B and the D claims, but it was by far the greatest for the less known claim. Again, the disclaimer did not change those ratings of perceived other benefits from this food. It was the same whether or not there was a disclaimer there.

When we looked at health importance, the more global rating, we only had a significant effect for the D level claim, lycopene, where it was more positive when there was a health claim versus not. But for the disclaimer, again, no effect. So, people didn't think it was better or worse for their diet as a result of there being a disclaimer there--no significant differences.

We also looked at what happened when you dropped out the "may" and we got what we sometimes refer to as a boomerang effect where people were more negative when they saw a claim that didn't have the "may" there, maybe because they are surprised to see it worded that way; that is not how they would have seen it before. Or, they may feel that this is just being a little too strong for what they feel comfortable with because we have seen that in the past. If people think a claim is inappropriate and is stated too strongly, trying to influence them, they may have a negative reaction rather than positive when they see that claim. The exception was the B level claim on scientific certainty and other benefits. People who saw that claim were more positive when the "may" was dropped out.

We compared across all the subjects the unqualified health claim statement versus the nutrient content claim. Again, in the past we found that, to the extent people already know about these things and many of the SSA claims are pretty familiar to people, if you just see a low fat claim or a low sodium claim you don't have to tell people why that is important. They already know. And, we did find out with the calcium that they already know about calcium so you would get the same results with just a calcium claim as you would with putting the health claim on there because they already see that as a signal of that relationship.

But we got strong impacts for the unfamiliar nutrients regarding scientific certainty and the relevant health benefit and health importance compared to just seeing the nutrient claim. Again, this is probably because they know less about it and so they needed a little additional information to tell them why it mattered.

For the full information, after reading the summary people did give stronger ratings to scientific certainty for the A and B level claims, A being the SSA type of claim, not that it actually was ever labeled as A. And, the negative summaries, the C and D claims, were not significantly different. So, although we were trying to make them more skeptical of those claims, I think just by virtue of the fact that they were reading so much about it they didn't become more skeptical and their ratings were the same as somebody's seeing an unqualified statement of that claim.

Just to summarize, do disclaimers work? Well, in our study and the disclaimer schemes we looked at, the text only disclaimers were unsuccessful at conveying different levels of scientific support. The report card schemes that gave an easy metric, the A, B, C, D that most people are familiar with was much more helpful but it also sometimes created some surprises, for example, when the B and C level claims were rated more positively than the unqualified or, in effect, A claims. So, because the As aren't there people sometimes think anything you say makes it more positive.

The health claim effects we saw are comparable to what has been found in the past. People tend to be more positive about the food product and disclaimers don't change that positive view that the health claim creates. They often have stronger effects for things that people know less about. So, if it is giving them some new information, some value added information, it has a stronger impact.

People's prior beliefs, of course, are important and if a claim is perceived as being too strong, stronger than it deserves to be, then it can cause a negative reaction. So, there is no guarantee. Even though health claims in general have a positive effect, if it isn't credible to consumers it could have this boomerang, and any information that is to enhance what people know, even when it is highly qualified, may increase their confidence even when you try not to.

The full report is available online. It has charts and more detail on all these things. That is a cumbersome address but it works. And, that is it.

[Applause]

MS. FRASER: Thank you, Brenda. She has done a miraculous job of reducing a very detailed and thorough report to a nice, concise presentation. So, I thank her for that. We have about ten minutes to take clarifying questions only, if you want her to explain something about the study or something that she said. If you do, if you would come to one of the two microphones--one is here and one is there, state your name and your organization if that is applicable, and pose your question to Brenda, and we will put her in the dunking booth and see if we can dunk her here. So, are there any questions?

[No response]

Great.

DR. DERBY: I will be around all day.

MS. FRASER: Please, step up to the mike.

MR. EMORD: [Microphone off; inaudible]--let me understand that the data you presented are not conclusive but [inaudible]--?

DR. DERBY: Well, it is never 100 percent conclusive. These are research results. These are consumer ratings. You never hear scientists saying something is absolute. What you look for is consistency across different studies, across different measures, the pattern of results more than any one question or one result.

MR. EMORD: [Microphone off; inaudible].

PARTICIPANT: Turn the microphone on. We can't hear you.

MS. FRASER: I think the mike is on. You might just need to move a little closer. No?

MR. EMORD: Now it is on. Well, my name is Jonathan Emord and I am with Emord Associates. I have a question for you. I imagine that when someone is asked a question in these surveys the people, of course, would come with preexisting knowledge necessarily from the popular press and other sources that may affect their general understanding of the nutrient-disease relationship or their opinion about it. Were any measures taken to exclude from the testing those who had certain levels of knowledge or understanding about the nutrient-disease relationship?

DR. DERBY: No. The way you control those kind of individual differences in an experiment is that you randomly assign people to a condition so that is a way to control their personal characteristics. Also, by having the control condition where there is no information and asking the same questions, you get a sense of what the prior knowledge is in the population because if it is a relationship people know more about you will get that in your means.

MR. MORTON: Did you assess any correlation between the extent of preexisting knowledge and the level of the qualified claim? In other words--let me rephrase the question, did the extent of your preexisting knowledge tend to cause one to produce a result that was more consistent with what was expected on your grading system?

DR. DERBY: We did ask them ahead of time for their familiarity. I don't think we have run the model with that variable in there but we could do it.

MR. MORTON: That would be interesting.

DR. DERBY: But given that you have randomly assigned people, you are kind of washing out what you would find otherwise. In a survey you would be able to do that but in something like this you have put them in different conditions in such a way that I think it would be harder to find that kind of effect.

MR. MORTON: Thank you.

MR. EARL: Robert Earl, with the Food Products Association. Brenda, thank you for your presentation. The question is as a follow-up in moving forward into the future. Can you describe--or perhaps this may be applicable to Steve or Alan as well, what are your plans for taking learnings from this research and looking at your future research projects on health claims, and whether you plan to do anything like that, and what some of the structures are--since I know you have some things in the pipeline related to foods versus substance and other types of things like that may help inform the situation?

MS. FRASER: I will ask Steve to come and talk about some of the research that is currently being planned. In terms of what we may be doing in follow-up of this specific research, that is in part what the purpose is of this conference today and the comments that we are receiving both orally and in writing to help us shape what are the additional research needs. But Steve can certainly talk about some of the studies we currently have planned and are working through getting cleared.

DR. BRADBARD: Just real quickly, we have a couple of studies that are, as you said, in the pipeline. Dr. Jordan Lin is the principal investigator for one of these studies, and this study is going to be looking at how consumers respond to claims for foods rather than for the substance that is believed to affect the health benefits. For example, how do consumers respond to a claim such as "yoghurt may reduce the risk of osteoporosis" versus "yoghurt which contains calcium may reduce the risk of osteoporosis?"

Another study that we have planned, and Dr. Conrad Choiniere is the principal investigator, and he is looking at some of the recent claims that FDA has allowed under its enforcement discretion, and we are going to be examining consumers' reactions to the actual claim language that has been allowed for omega-3, for mono unsaturated fatty acids, and for green tea for both conventional foods and dietary supplements.

MS. FRASER: Thank you. You will have the last question and then we can take our 15-minute break.

DR. MAGNUSON: Berna Magnuson, Burdock Group. May I ask a question of clarification from Louisa Nickerson?

MS. FRASER: We really are limiting the clarifying questions just because this is more focused on the research. I would say see Louisa at the break. I think that would be a better use of our collective time. Part of it is that I am trying to see if there are others who want to speak later, if we can save as much time as possible.

With that, I have 10:15 by my watch. We will resume promptly in 15 minutes. Again, the restrooms are at the top of the stairs. Don't bring the food and drink back in here, but please do eat what is available outside. Thank you very much.

[Brief recess]

MS. FRASER: Please take your seats. We will get started or resume, as the case may be. It gives me great pleasure, and we are very thankful to have Dr. Pauline Ippolito with us, the Associate Director from the Bureau of Economics at the Federal Trade Commission, who will next present the results of FTC's research findings. Thank you.

Federal Trade Commission Research

[see presentation slides for Dr. Ippolito]

DR. IPPOLITO: Thank you. I sort of feel like I am in a pit here. People may start throwing things! I should begin by saying that the opinions I express today are my own and not official agency positions.

I assume most people in this audience know this but let me begin with just a little bit of background. The FTC is involved in health claim issues because we regulate food advertising, not food labeling. So, obviously, if you are going to market your product as a healthy product it is probably going to be on the label and in the ad.

So, we have also been doing research on how to communicate on certain science, and how consumers react to this science, and how we would measure it. So, what I am going to do today is present just a few selected results from that research which has been going on for some time and which has been led by Dennis Murphy of our staff in the Economics Bureau.

The FTC does not have the kind of specific regulations that the FDA has. We govern all advertising to consumers virtually, except for prescription drugs and a few other things. So, we have general principles that apply to ads made for all kinds of products. Those are basically described in two major statements in policy, called the deception statement and the advertising substantiation statement, at the FTC. But basically they say that claims must be truthful and not misleading, and firms must have a reasonable basis for any material claim that a substantial number of reasonable consumers would take from an ad. I am paraphrasing there.

So, when it comes to claims based on science, the issue is what claim can you make given the state of the science? So, there are two things in play, how strong is the science and what can you communicate to consumers about that science? That is how we get into the issue.

We have been doing research on qualified claims. Two key things that we are interested in are can you communicate uncertain science to consumers, or do consumers assume that the science is basically well established when they see a claim? And, can you communicate different levels of science? As we had begun this process, we were thinking about science that was at least weight of the evidence, that is, taking the body of science as a whole, the weight of the evidence was supportive of the hypothesis of the claim that was being made. With recent court decisions we have had to look deeper into science. Could we communicate less than the weight of the evidence? As I said, we have been doing this for some time.

The results I am going to talk about today are all based on this template, or pretty much all based on this template. This is a fictitious product, as you can see. It is basically an antioxidant supplement--same picture; same layout. What changes is the text in various treatments. So, we would modify the text and see whether it matters in the consumer measures.

This particular ad--let me go back a minute. We start out to say what do we know about antioxidants in cancers. "Scientists have known for some time about the special health benefits of fruits and vegetables that are rich in antioxidant vitamins like vitamins A, C and E. Eating plenty of these foods can reduce the risk of certain kinds of cancer."

Now, this is the part that changes. "Some medical studies are now suggesting that supplements containing these same antioxidant vitamins may also reduce the risk of cancer." What this means for you--"it looks promising but scientists won't be sure until longer-term research is completed." Then it goes on to, you know, "eat a good diet."

This is what I am going to label the mildly qualified claim. The qualification is all in positive terms. "Studies are now suggesting;" "it looks promising;" won't be sure until we are done with the rest of the research. So, that is the part that is the mildly qualified claim.

To have a control we did a proof claim, what I am going to label a proof claim. This is a very strong claim. It says "scientists have now proven that supplements containing these same antioxidant vitamins also reduce the risk of cancer. It is a fact now." That is not an SSA claim.

[Laughter]

Now, we deliberately made it strong, maybe stronger than we would like to see, so that we would have a strong anchor against which to test the qualifications. The second claim is the mildly qualified claim which we just talked about. Then we have a qualified claim, what I am calling a qualified claim, which is a little bit stronger. The top paragraph is the same as the mildly qualified but then it says "it is too early to tell for sure. Some studies have failed to show that these vitamins protect against cancer. Longer-term research is needed." So, it is a bit more negative. "It is too early to tell;" "failed"--bad word, failed. Ad copy people don't like negative words. "Longer-term research is needed." In the earlier study "it looks promising," a positive word; "won't be sure until it is completed." Okay? So, it is a tone distinction, a little more negative. Does it really matter to consumers?

Those were our earliest tests. When the court cases came down and we were looking at the prospect of more qualified science and allowing claims for more qualified science we tried to come up with a more highly qualified disclosure. So, that is what I have here. "Some science suggests" is still the same but then there is a box disclaimer. It says, "there is much scientific debate about whether antioxidant vitamin supplements reduce the risk of some kinds of cancer. Most studies have failed to show that these vitamin supplements reduce the risk of cancer." So, a much stronger qualification, hopefully, reflecting weaker science. Also, set off in a box, more like a warning than traditional ad copy, again, we thought is a stronger qualification.

Let me give you the questions and let you look at the results while I give you the question. Based on what the ad says or suggests, how sure are scientists about whether taking antioxidant vitamin supplements will reduce the risk of certain kinds of cancer? So, it is asking specifically about supplements. It is asking based on what the ad says or suggests, how sure are scientists.

We asked people to rate on a five-point scale from unsure, which was 1, to sure, 5. These are the mean response rates for the different treatment conditions. The first thing is we are getting the kind of ordered response that you would like to see. As we went from proof to mildly qualified to qualified to highly qualified you get a systematic reduction in the mean response on how sure are scientists. So, that is good. All the differences are statistically significant.

Now, a proof claim which said it was proven; it is fact, consumers are rating as about somewhat sure. So, there is clearly some discounting going on. They are not taking the literal statement as definitive. That is not an unusual result. Throughout the ad literature, consumers are skeptical of ad claims. You know, it is a selling message. Consumers know that. There is a certain amount of discounting that goes on. We are seeing that here but we are getting the systematic pattern that we were hoping to get.

On the highly qualified box claimer we are getting a mean rating that is just beneath the mid-point of the scale. So, the real question in terms of policy in terms of deception is, is that enough. Is that a low enough rating for a C or D level claim? I don't know the answer to that.

Now, one concern we had was, well, this is a supplement. Maybe supplements are different. At the time, antioxidants were not that well known and so maybe we are getting some discounting of the proof claim because of those features. It is a supplement and it is an unknown relationship.

So, we also tested that proof language on a claim for a cheese product, a low fat cheese product and the calcium osteoporosis claim which is a very well-known claim, and those are the results for a proof claim for that treatment, and there isn't a significant difference between those two. So, there is some discounting going on just because it is an ad treatment probably.

Now, these are mean results. As a matter of enforcement, when we bring an ad case what we have to establish is that a significant number of people were deceived. So, you are really looking at how many people got a deceptive message. So, another way to look at this data is to look at the right tail, if you will. How many people said they were sure, the scientists were sure, which might be a deceptive message if you aren't in the proof case. This represents those results.

So, how sure are scientists? The percentage of consumers who responded that they were sure was 58 percent for the proof claim, so another measure that not everybody is taking the literal language literally. Then it drops off quite quickly so that the mildly qualified claim is only 22 percent being sure, and then 10 percent and 15 percent and 5 percent. So, it does appear that most people got the message that the science wasn't sure once we introduced even the mildly qualified language. The cheese result comes in pretty comparably as before.

While we were doing all of this, FDA came into the business and they started proposing language in their consents. So, we had this scheme and we decided to test some of their language in our treatment approach. We were also a bit concerned when we had set up this five-point scale. There is a well-known phenomenon in the literature that a lot of consumers don't like going to the extreme boxes and so, you know, having only a five-point scale was something we wanted to look at. So, we did some additional testing using a seven-point scale, the same kind of idea, but we went from not at all certain, 1, to very certain, 7.

We wanted to test the FDA language. Since we were doing antioxidants, we decided to try to test the C language. So, this is some "scientific evidence suggests that consumption may reduce the risk of certain kinds of cancer. However, FDA has determined that this evidence is limited and not conclusive." So, that is the point-counterpoint approach.

Then, the report card language, same beginning, "FDA evaluated the scientific evidence and gave it a C rating, based on a scale from A, strongest evidence, to D, weakest evidence." To have a measure back to our older work we also tested the highly qualified box disclaimer that I showed you earlier.

First, in terms of the metric, the highly qualified box disclaimer is again coming out just a bit below the mid-point of the scale so in that range it doesn't seem to be affecting where it is positioned. The two FDA approaches that we tested are coming in above our box disclaimer, which is where you would expect them to be if it was a C kind of disclosure.

When you look at how many people are being deceived given that we have a seven-point scale, what I am showing you here is people who responded 5, 6 or 7 so the top part of the scale. We are getting quite a number of people who are telling us that the evidence is that strong, which might be a reason for concern if, you know, we were looking at C or D type science.

Let me give you a little bit more detail here. This is the highly qualified box disclaimer. This is the whole distribution of responses. First the good news is that very few people are responding 6 and 7 with that highly qualified box disclaimer. That was a highly qualified disclaimer. I would be surprised if very many advertisers would want to run that kind of disclaimer. What is a bit disturbing is that quite a few people, a quarter of the sample, are saying somewhat certain at 5. So, that might be a reason for concern.

More broadly, it is clear we don't have a very finely honed message here. We are getting a very disperse distribution, almost uniform if you chop off the right tail. So, it is clear we aren't communicating a tight message to most people. We are getting a lot of variation in our response.

The report card disclosure did a little bit better in the sense that more of the mass moves to the middle. We are beginning to look like a normal distribution here. So, C is communicating middle to people, it appears. The language is more all over the map, as it was in our case, though very few people are doing 6 and 7 again.

One of the concerns we had is how much of this is an ad effect? You know, the fact that we are getting these very dispersed reactions because maybe people have different reactions to advertising. So, to test that we tried, in a different set of experiments, a lycopene fact sheet which was completely outside an ad context. It was, you know, black and white; lots of words on paper; an information sheet as it was called. You know, it was trying to describe the state of the science. So, "a number of studies have found that people who eat diets rich in tomatoes and tomato products tend to have fewer heart attacks and other heart problems. Scientists have also studied whether some of this benefit may be due to lycopene, which is a nutrient found mostly in tomatoes. So far, we have learned that people with heart disease have less lycopene in their bodies than heart-healthy people do. Some studies have reported beneficial effects on cholesterol in the blood when people take lycopene supplements. But other studies have not found any benefits. At present there are no long-term studies of whether people who take lycopene will actually lower their risk of having a heart attack. Wo we do not know whether there is any benefit from taking lycopene supplements. Carefully controlled and long-term clinical studies will be needed to answer this question."

So, it was an attempt to be neutral, accurately describe the science at the time--this was a few years ago--outside an ad context. We are getting the same sort of very dispersed kind of reaction. We aren't getting a tightly honed message. People aren't used to these kinds of messages. Maybe they react very differently to them.

So, from our point of view it wasn't the ad effect that was driving those dispersed distributions; it was more the content of the message. That was a sample of some of the things we have done.

So, what are some conclusions I would like to leave you with? Well, qualifying language does change the message consumers take away in our tests. By that, I mean that the mean consumer perceptions of scientific certainty fall with stronger language. That was this picture.

So, this is a somewhat different result than the FDA treatment in that language alone here affected the take-away but it was very different language than the FDA tested, and it was designed to try to very carefully ratchet up the qualification. In both agency's cases we didn't have top-notch ad copy people doing these tests, designing these messages.

So, I think in terms of future research this is potentially an area we could do some work on, we could use some work on. Are there words that will work is a different question than our words seem to work; their words didn't seem to work. You know, there is room for discussion there on what might work and are there better words. Our copy is long-winded. It was designed with ads in mind. It was designed to really try to do a basic test early on. It is probably not what most people want to use and certainly not on a front label. So, I think this is an area where we could use some more work.

We found pretty strong evidence that even with our very strong claim--proved; it is a fact--we are getting significant discounting. As we talk about qualified claims, remember what the big policy issue is. In an ideal world we would like consumers to know where there is very strong scientific support for actions they might take as opposed to areas where there is good science but it is not conclusive, but they might want to consider changing their behaviors, and knowing the difference between those two.

We have all been focused on how do you qualify, but the other side of it is how do you make the A claims stronger? And, that is something that could be thought about. Is there a way to essentially certify the class A science somehow that is easy, that is quick, that communicates that this is really some solid science and there isn't much debate about it anymore, as opposed to the other kinds of claims that might be made? So, that is another area that I think we could possibly think about a little bit more seriously.

Then, the third thing that we don't know what to do with, other than try and bring in some professionals maybe, we are getting very dispersed distributions. We don't have tightly honed messages in the ad copies that we did, for sure. So, you are left with both types of error. You have people who are overstating the certainty of the science, but you also seem to have quite a few people who are understating the certainty of the science when you use the kind of language that we have used. Both types of errors really are problematic in terms of trying to get people to absorb the science that is out there and to change their behaviors in ways that are productive for health. We would like to get cleaner, clearer messages to folks. So, how we do that I think is an important part of what we all should be thinking about.

We do have a couple of public reports. That is on the last slide of the handout. They are both on our web site I believe and we would be happy to mail copies to anybody. Thank you.

[Applause]

MS. FRASER: Thank you. We probably have about five minutes for any clarifying questions for Dr. Ippolito. Please step to the mike and state your name and affiliation.

DR. MURPHY: Dennis Murphy, Federal Trade Commission. I just wanted to mention one fact that will maybe pull together the FDA study and the FTC study a little. We did have a control condition for the ACE vitamin product where we asked people who got no claim whatever what their prior assumption--what do you believe about the state of the science. What was interesting was that that was the lowest score. So, people were coming in with very low prior beliefs about the science, which I was very surprised about. So, the disclaimers are not fighting against a very strong prior belief that there is a certain science. Correct me if I am wrong, Alan, but I believe this is also the consistent result you were finding when people saw a no claim condition. They were quite skeptical.

MS. FRASER: Thank you, Dennis.

MR. EMORD: I guess I would ask you or state the same thing I stated before, that when you deal with studies you really look at probabilities of outcome and not conclusive evidence.

DR. IPPOLITO: Right, you are looking at testing a hypothesis.

MR. EMORD: So the FDA and the FTC do the same thing.

DR. IPPOLITO: Sure. Let me throw one other hypothesis out there. In terms of the FDA finding that the SSA claim actually scored lower than the B level claims in many of their tests, one difference between the two claims is that in the B and other claims you are talking about the science in the claim. In the A level claim there is no discussion of science. There is no mention that there might be scientific support for this claim; it is just a statement. So, then you ask consumers how strong is the science or, you know, what do scientists think about this science you have given them no basis for answering that question, which either may be an artifact of the measurement or it may be that you haven't persuaded them that there is solid scientific support.

So, again, it goes to this issue of looking at the A claims, as well as the qualifying claims, to try to make sure that people are getting the message that A claims really are the A claims.

MS. FRASER: Thank you very much. We next are pleased to have Wendy Reinhardt Kapsak, from the International Food Information Council Foundation do a presentation of industry's research in this area.

International Food Information Council Foundation's Research

[see presentation slides for Ms. Kapsak]

MS. KAPSAK: Hi. I am very thankful to be here today and would just like to tell all the researchers here that I have had an opportunity to speak to many of you and I am so grateful to you for sharing your findings and insights with us, and I would also like to particularly thank FDA for inviting us here today and for really assisting us with several aspects of this project.

Today I am going to talk about a few things that some of you may or may not be familiar with at the Foundation, just to clarify a little bit about who we are and where we stand, and also let consumers have a say. So, we will present a lot of the consumer research not just on qualified claims but also other kinds of health claims and then label statements. Then I will go ahead and summarize the findings.

The International Food Information Council is a non-profit organization, located here, in Washington, D.C. The main mission of it is to communicate science-based information on a wide variety of food safety and nutrition topics basically to anyone who is on the front line communicating with consumers every single day so health professionals, media, educators, government officials--again, those people that are focused on communicating with consumers every day. We are funded by the broad-based food, agricultural and beverage industry. What makes us really unique in Washington is that we are not involved in any kind of lobbying or advocacy work so in that sense we are quite unique.

IFIC Foundation is the educational arm of IFIC and this particular research project actually came through the Foundation. So, a lot of the research that IFIC does is actually on the IFIC Foundations' web site, including an executive summary of this research and there are also some copies that may be being passed around and are in the back as well. There is also a Spanish portion on the web site. Also, other educational initiatives, including the Food Insight Newsletter and many different other IFIC Foundation publications can be found on the web site that are truly the educational arm of IFIC.

As we thought about how to put this project together, and we did consult with FDA at that time just because it was stated that there would be consumer research being done, we wanted to find out could we be helpful; could there be anything that IFIC could add to this discussion. So, in that, we put together several different objectives following some of these discussions, particularly with FDA. We wanted to measure consumer reaction to the FDA proposed four levels or the schematic of health claims on the basis of the strength of the scientific evidence. How do consumer perceive evidence based on that particular schematic.

But this is the part where FDA looked at a number of different outcome measures and we did as well. One that we have in common is the strength of the scientific evidence where we kind of went above and beyond in that we wanted to also look at the overall perceived healthfulness of the product; how does the claim that is being presented to consumers impact the healthfulness of the product; the perception of the product quality; and the perception of the product safety. Then, would the claim's presence have an impact on purchasing intent.

We also wanted to determine if consumers were able to differentiate between dietary guidance and health claims, and there were a few statements already put up in some of the introductory presentations. Then, we also wanted to examine the impact of structure/function claims and some alternative language to what was being proposed at the time, and how that then compared to consumers' perceptions between those kinds of claims and the current unqualified claims.

We commissioned Cogent Research, out of Cambridge, Massachusetts for a lot of this research and Cogent, IFIC and FDA all worked together in terms of the survey design and then later on some additional methodology and some additional analysis. I don't want to get into too much--I am more than happy to answer any questions about how we put the survey and everything together. This has been a crazy project and I am surprised I am still living! I mean, really!

[Laughter]

I mean, I am going to spare you a lot of the hurt that I have already endured!

[Laughter]

This is actually one of the largest consumer research projects actually completed at the IFIC Foundation and I don't know if we will ever do another project this large, based on what I just said. It had over 5,600 participants. Most of our quantitative surveys are around 1,000 folks and this was about 5,600 participants. It was a web-based survey and the data was weighted to match the 2001 U.S. Census data.

There were three products that we tested and two of these were similar to the products that FDA tested, orange juice and then the relationship, calcium and osteoporosis; pasta sauce, lycopene and cancer. These were the two that were similar. Then, this kind of new one which was breakfast cereal and the relationship was trilinium and diabetes. Is anyone here familiar with this? But five percent of the U.S. population is familiar with this relationship.

[Laughter]

So, that speaks to a little bit to it right there. And we did test the awareness of the other two, by the way. We also looked at making sure that definitely an aspect of this project, or a part of this project was really moving in parallel with the FDA study as we looked at some of the same kinds of tested formats, claim formats, and Brenda mentioned some of those.

The four that were consistent between the IFIC research and FDA were the report card graphic and I will show an example of that graphically so that you can see that; then the report card text where we take the letter grade and essentially put it into the language itself; then other language only claims, claims that don't mention any kind of letter grade at all and these were embedded point-counterpoint.

To show you a little bit about how those looked like, we actually tried, to the best of our ability, to make our report graphic claim very similar to that of FDA's. So, on the left-hand side you see the actual claim and then you see this check-box idea over to the right. So, that is an example. Looking back, you know, it is always in retrospect, our control actually did include a nutrient content claim. So, I just wanted to make it very clear up front that that was a slight difference.

The report card text, here you see that the claim language, very long claim language, is actually listed at the bottom of the label here, and you can see that this component, in this case lycopene, may reduce the risk of disease and then, again, the language that the FDA was also testing. FDA evaluated the scientific evidence and gave it a B rating based on a scale of A, strongest evidence to D, weakest evidence. So, the graphic included just what it says, a graphic or a check box, and then the text took the letter and put it into the language of the claim.

Then we have truly language only claims that don't mention a letter grade at all embedded where the qualifier essentially is given up front, in the very beginning, promising but not conclusive evidence suggests.

Or then, point-counterpoint where it gives the actual statement of the diet-disease relationship but then there is a second statement that says the scientific evidence is promising but not conclusive.

I mentioned that we also looked at structure/ function language, and we did that also for the three products that I mentioned before. So, the calcium and orange juice example also helps promote bone health; lycopene helps maintain prostate health. That was on the pasta sauce example. Then the breakfast cereal, trilinium helps maintain a health blood sugar level. I did go ahead and run some of these by someone over at FDA, just to see are you okay with these in terms of testing them and they said they seemed pretty reasonable for what we were doing. We also tested some dietary guidance statements and I will go ahead an mention that in a later part of the presentation.

Now I am going to just go ahead and briefly just go over the top line findings. We repeat them at the end but we will also look at them individually. First off, this was mentioned in one of the earlier presentations, it is really, really difficult for us to essentially try to only look at one facet of what could be impacting consumers' let's say, critical evaluation of a particular claim. So, the claim type, whether we are talking about embedded or graphic, whichever, claim level, A through D in this case, and/or the perceptions of the raw product--do I feel orange juice is health? Is that a healthful product? Or, am I aware, for example, that lycopene may somehow impact prostate health? All of these things, among many others which we are going to hear this afternoon, can really impact how a consumer perceives or is making a critical judgment about a claim and its believability.

So, it is really difficult. When we are trying to look at these things in isolation it is really difficult to make some of those points that have been made really conclusive results about any particular one of these outcome measures.

But, to move forward, consumers have difficulty distinguishing among four levels of scientific evidence, especially with language only claims. Again, that is that point-counterpoint or embedded claim where there is no mention of a letter grade whatsoever. So, consumers have difficulty distinguishing among those four levels.

Thirdly, consumers can distinguish among four levels using a report card graphic, and we might expect that and we will get into that a little bit more, just based on what we already know about what A, B and C and D have meant to us in our education. But, unfortunately, with this particular schematic or system or claim format we really did see some negative effects observed with consumer perception of product healthfulness, quality, safety and purchase intent. This was seen in various instances with the report graphic, as well as the report card text. So, this idea of trying essentially in this framework to use a really simple symbol or way of trying to express the level of science, while it was helpful in some ways, was actually misleading consumers in other ways.

Finally, consumers rate the scientific evidence and other attributes of a product containing an unqualified claim similar to that of products containing either a structure/function claim or a dietary guidance statement. So, basically they were rating these particular statements very similarly with regard to strength of the scientific evidence.

So, we mentioned the first finding and I will let others speak more to this in the afternoon. But we can't look at each of these things in isolation. In fact, claim type, claim level, perception of the overall product and then the awareness of the diet-disease relationship, among many others, all impact critical evaluation of a claim or label statement, or perhaps then also ad copy.

So, now as we get more into the meaty part of our findings, we will go ahead and look at them individually. Consumers have difficulty distinguishing among four levels of scientific evidence, especially with language only claims. This is the point-counterpoint.

What we did in this situation was we actually had consumers do what we were calling a card sort exercise. So, we gave them four different kinds of statements. What is important to point out here, and for the rest of the discussion, is that in this particular IFIC study we did look at A level claims in this study. So, it is important to note here, and we can make some judgments based on what we found in the data about how consumers may be looking at A or unqualified claims in relationship to all these other levels.

So, in this case they were given four statements, and what they were asked to do is put them in order, one through four. One would be the strong evidence and would be the little evidence. Then, you can kind of see the language that was included in the claim statement underneath each of these cards.

So, the first one, we would have hoped people would have placed that with the actual current unqualified claim--calcium may reduce the risk of osteoporosis, let's say. Promising but not conclusive would be considered the B. C, limited and not conclusive. Then, very limited and preliminary. So, we would have hoped that people would be able, looking at the language, critically make a judgment to place these in the correct order.

What we found is that the majority of the consumers incorrectly placed claims as to the level of scientific evidence. If you look at responses given by statement, you will see a stair-wise process kind of happening with those marked in the red box. So, if we look at the unqualified claim, about 36 percent of those folks put the A claim where it should have been, in first place, giving it a number one essentially. But then if you look at the right-hand side you see that 64 percent of consumers actually placed it in the wrong place. They didn't necessarily put it in D but they could have put it in C or B. You see the same thing happening with B, 39 percent put it in the correct place, the B place, but then 61 percent put it in the incorrect place.

So, the idea is that in giving the consumers a card sort exercise, thinking that they would critically look at the language and then make some inference based on that language, they didn't do what we thought. A majority of them placed it incorrectly. So, 78 percent of consumers actually cannot correctly sort these four levels. Nearly a quarter got it right so 22 percent got it right.

We then asked a question, hey, do you think this was easy? Was this hard to do? You know, we were kind of trying to get at did they think this was easy. Only one-fourth of consumers felt it was easy. Out of those, one-fourth, only 32 percent of those actually got it right.

I also just want to make a statement that we did ask a battery of questions in the beginning to determine, you know, would someone would consider themselves to be what we would call more health active, someone who says, yeah, I take a really strong interest in searching out this kind of information; my family and friends actually look to me for nutrition information--those type of questions. Those health active folks, you know, we thought maybe they would be actually suited for this kind of exercise and, by having the knowledge that they have, perhaps they would be a bit more critical about looking at the language included in these statements.

What we actually found is that they were not. So, the health active consumers were no more likely than what we would consider moderate or just very general active consumers in terms of putting these in the correct order. So, by someone thinking that they were more capable or had more information, those folks were no more able to do this exercise correctly than those who were not health active.

Now we are going to move on and say, all right, what happens if we add a report card or a graphic here? So, consumers can distinguish among four levels of scientific evidence using a report card graphic. The green line here--this is very similar to how Brenda presented here data--this is the report card graphic, again looking at that check box. What we found is that consumers can truly distinguish among four levels of scientific evidence using a report card graphic. We would expect this in the sense that people know, for example, that A is better than B, B is better than D in grade school, and so on. So, this isn't necessarily a surprise.

When we looked at the report card text--that is when we go ahead an pull in the actual A, B or C and pull it into a statement within the claim that says FDA evaluated this claim and gave it a B rating on a scale of A to D--what we see happening here, and in this case we did test a claim, is that a two-tier approach or a two-tier system seems to be happening. While we would think that consumers read the claim and are actually thinking A, B, C and D, in fact, they actually clump or put together A and B and push down C and D. So, in this case A and B are not distinguished from each other, and C and D are not distinguished from each other. But there is a difference between these two levels, those two that I mentioned, A/B, C/D.

Now, what happens with point-counterpoint? In these two language only formats, point-counterpoint and embedded, again, we used the current unqualified language for the A level and we did test that. But what we see here is a much weaker two-tiered system essentially emerging. So, in this case B is distinguished from C and D but C and D are not distinguished from each other. Okay? So, the two levels here are B and then the second level is C/D. Then for the embedded we found very similar to what FDA found. Really consumers did not distinguish among these levels at all.

So, what about the A claim? How is the A current unqualified language viewed, for example, within report card text or graphic? It is actually viewed as being a C level. So, if you still wanted to somehow keep a record card in terms of qualified, some kind of check box or using letter grades B, C and D, what we find is that if we keep A level language the same it will be viewed very similar to a C level claim.

So, consumers can distinguish among four levels of science using a report card graphic and then we saw how the others fared. But then with this graphic we saw negative consequences observed in consumer perception of product safety, quality and healthfulness at some lower levels.

A lot of these slides are going to look very similar. Really what you need to know--and all of this actually is included in the executive summary. If you go back to your office, or whatever, there is a section for you to click so you can keep some of these major findings right there. But the statements at the top of these slides really do indicate where there are statistically significant differences.

What is important I think to note here, in my opinion, is that while I am listing out for you where the statistically significant differences are, in some instances it is also important to note where there were not statistically significant differences as well.

The C report card text and D report graphic in some instances conveyed less helpfulness than structure/ function and several B kinds of claims. D report graphic and text--and you will see those denoted by the green on the graph--convey less quality than structure/function and alternative B, and I will explain what alternative B is at another time. And, D report card graphic also conveys less quality than report card text A. So, that is a nice, clear piece right there. With safety, we see that C report card text conveys less safety than some A and B claims and also the structure/function.

I would really like to look at this data again to see if we remove some of the different statements--you know, what is interesting here is that we don't see the same thing happening in every single instance, but the point is that we are seeing enough to say, you know, there is something here. In some ways, this is really impacting consumers' perceptions of these particular product attributes in a very negative way.

I will explain, and Dr. Bradbard mentioned a little bit about focus group data and while we can't make this inference for the whole U.S. population I do think that this is something that really can drive the point home, that we asked consumers given two products, equal in all other ways but on product 1 you saw an A level claim and on product 2 you saw no claim, consumers stated that they would choose the product with the A on it.

Second scenario, two products, the same except for the statement that was on them. One product contained nothing on it, no statements of diet-disease relationship; the second product contained a statement but it also connected that D to it. They said I would pick the one with nothing on it.

Again, that was a focus group finding but it kind of just drives this point home that this kind of way or simplifying the scientific evidence can actually negatively impact consumers and actually mislead them. We saw also the same thing, that consumers are actually less likely to purchase a product with a D report card text claim than those who saw several different other kind of higher level claims.

This is the last finding, consumers rate the scientific evidence and other attributes of a product containing an unqualified claim similar to those products containing either a structure/function statement or a dietary-guided statement.

FDA looked at this data as well, what if we take the word "may" away from the current unqualified claim, and would that have any impact? That was actually one of the questions that was being asked. We actually found that consumers did not perceive a difference here in terms of scientific evidence. I think, you know, at various focus groups we found similar things. Sometimes consumers actually like seeing the word "may" and sometimes they don't. They think that is enough of a qualifier. So, in this instance there was no difference for us when we compared consumers' responses regarding the strength of scientific evidence between claims that contained the word "may" and those that actually deleted it or removed it.

We also compared the rating of scientific evidence for structure/function claims. In this case we actually found that consumers do, in fact, rate structure/function claims very similar to current unqualified claims whether they would include "may" or not.

So, to Dr. Ippolito's point before, does this, for example, state, as I said earlier, that current unqualified language in a report card format would really be considered a C level claim, and now we see here that there is designation between structure/function and current unqualified language. Does this beg the question should we be looking actually at unqualified claims and pushing up the language included in there just by the very fact that we may even talk about science in that claim as a change?

Now I want to move on to dietary guidance statements. We did look at different kinds of products here. Orange juice vitamin C and cancer was the relationship; yoghurt calcium and osteoporosis; and then tuna omega-3 and heart disease. I am going to go through a little bit about what these statements were like just for the orange juice, cancer, vitamin C diet-disease relationship, but if you look at the executive summary you will see all of the different kinds of statements listed there.

The first one--this was actually an alternative to what the current dietary guidance statement is for fruits and vegetables, the current dietary guidance statement that is made in conjunction with FDA and then the National Cancer Institute. So, based on some of the earlier presentations, we wanted to look at what about a more general statement around dietary guidance? So, the alternative dietary guidance statement we tested was "eat five fruits and vegetables a day for good health." Then the current dietary guidance statement for fruits and vegetables, "diets rich in fruits and vegetables may reduce the risk of some types of cancer and other chronic diseases." So, there is a real big difference here in that, in fact, the dietary guidance statement is actually mentioning not overall general health but talking about risk reduction for certain chronic diseases.

Then there was a question in some earlier FDA communication, what if we then also included a dietary guidance statement in the product? So, "diets rich in fruits and vegetables, including orange juice," in this case, "may reduce the risk of some types of cancer and other chronic disease?" So, we included the product here.

Then we also wanted to compare this to current unqualified claims for a similar kind of relationship. So, here is the current unqualified claim. Then we also wanted to pair the current unqualified claim with the product. So, in two instances we included product, one with the dietary guidance statement, one with the current unqualified language. We did a very similar thing then for the other two products as well.

On the left-hand side you will see a straight line. Basically, what that is letting us know is that out of those five statements that we tested--alternative dietary guidance, the current dietary guidance, dietary guidance with product, unqualified claim, unqualified claim with mention of a product--consumers did not rate any of these different based on the scientific evidence supporting the claim. So, they were looking at them very similarly. They weren't looking at them differently.

Then, just to pull back in some of the data around current unqualified language, around "may" or no "may" and then also structure/function, we really do not see consumers being able to or wanting to distinguish among these kind of statements, dietary guidance, unqualified claims or structure/function. They are viewing them very similarly in terms of the scientific evidence.

So, this is a repeat and I will go through it quickly, but just to kind of ground us back to where we began, the claim type level perception of the product, awareness of the nutrient, in addition to a bunch of other things, actually plan a role in consumers critically making a judgment about various kinds of claims.

Consumers have difficulty distinguishing among four levels of scientific evidence, essentially with language only claims, point-counterpoint and embedded. When we add a graphic in there, that seems to help consumers in terms of distinguishing among scientific evidence within those four levels but, unfortunately, we see some negative effects happening at lower level grades in terms of other product attributes--healthfulness, quality, safety and also purchase intent.

If we look at current unqualified language, consumers really rate the scientific evidence behind those, or that diet-disease relationship, very similarly across three kinds of statements, unqualified, structure/function and dietary guidance.

So, in summary, consumers had difficulty sorting out strength of scientific evidence associated with various claims, regardless of the type. We found in some other kinds of research, a lot of focus group research, that perhaps consumers are really searching out some simpler language and positive language, such as we find in structure/function. Perhaps, to Dr. Ippolito's point, maybe we haven't found the right language for health claims just yet.

Then, unintended effects were observed when we tried to simplify with report cards. We saw some unintended effects. So, while it can help consumers perhaps look at science, it has some unintended effects which can actually mislead consumers about other product attributes.

Finally, of course, further research could be definitely--we could definitely use that to find out the ideal number of levels that perhaps could increase consumers' ability to distinguish. We know from our research that four levels in this instance is likely too many, but what we don't know is if three or two--what would truly be the right number of levels. Also, the terminology or language that consumers would find most helpful and motivating in terms of making healthful choices.

So, thank you so much for your attention. I really appreciate it.

[Applause]

MS. FRASER: I want to thank Wendy and if I could get her to stay for five minutes of so, if there are any clarifying questions, please step up to one of the microphones in the aisle and state your name and affiliation. We have a little bit of time and I think we might be able to get to the next presentation before lunch and that will give us some more time in the afternoon for comments.

MR. EMORD: Hi, Jonathan Emord again. How are you?

MS. KAPSAK: Good. How are you?

MR. EMORD: From all these presentations, it occurs to me that there is sort of a pre-supposition that FDA placement of a claim in a particular category, A, B, C or D, is an appropriate supposition and that if there is a deviation from it, for example, on the part of an individual consumer that that has not been ascertained and evaluated. In other words, if you were to ask a scientist who was well versed in the particular nutrient-disease relationship, he might well disagree or she might well disagree with the FDA's placement. How would that then affect a response? Alternatively, if someone was completely unaware of the scientific information in a particular area, they might well not have any specific regard or respect for the FDA's scientific assessment. In other words, the question is what, if anything, was done to ascertain the extent of knowledge, education and training of the participants and how that would affect their assessment of the rating system or responses to the questions?

MS. KAPSAK: I think you might have asked a similar question of the other research too, but basically we haven't cut it like that per se. The only thing we did that would get to that was that card sort exercise and, again, it is important to note that we are talking about general consumers here. So, what we did with general, broad consumers is we did do that card sort exercise, regardless of all those knowledge things which we did look at as well to find out, for example in this case, could health active, those who we would expect to make a better, more critical judgment--maybe not at the Ph.D., but make a more critical judgment, were they able to put these in the correct order, and really what you are saying is critically evaluate, and they were not.

But, certainly, after we made them look at all these claims, we did ask them how aware were you, and we did do some awareness and also some demographics that would include obviously what you are asking in terms of education and things like that. So, we could certainly attempt to cut the research that way but, again, I think what is important is whatever happens, I mean, we are not going to be doing it just for scientists or just for highly educated people. I mean, we are going to be doing it for all Americans so in this case that is what we were looking at.

MR. EMORD: Another confounding problem that I think may arise, and I don't know whether this was taken care of in the study, is people's preexisting opinion of the Food and Drug Administration. I mean, there is negative publicity about FDA, quite a bit of it in the last five years; hearings before Congress, and so forth, that are adverse. And, I would think that the average consumer who may be aware of that, it may affect their judgment as to whether or not when those statements say the FDA has determined this, whether that affects them. Was there any effort to ascertain what a consumer's perception is of the claim based on any preexisting bias for or against the Food and Drug Administration?

MS. KAPSAK: I have to look back at our data during the break, but we did ask consumers in focus groups a little bit about what you are saying, what their opinion was, who do they trust, and that kind of thing. Basically, what we found is that there is overwhelming trust in the FDA. In fact, they felt that FDA wouldn't allow certain kinds of claims actually to be on labels. So, there was actually a lot of trust from the consumers in those focus groups, and I will look to see if we asked the question.

MR. EMORD: Were those focus groups inside or outside the Beltway?

MS. KAPSAK: Outside, and there were no FDA-ers in them either.

DR. MURPHY: Hi. Dennis Murphy again. First of all, I want to congratulate you for what is an extremely extensive, ambitious and productive study. I am very jealous and I want to encourage you not to lower your ambition--

MS. KAPSAK: Please don't be jealous!

[Laughter]

DR. MURPHY: Jealous of your budget. My clarification goes to these negative effects you were finding for the report card scheme, kind of spill-over effects, say, on product safety. All the comparisons you mentioned were between, across different schemes, whether report card C was lower in safety, say, than an embedded B text. Of course, we are only going to end up with one scheme presumably. So, it seems to me that the more relevant question is did you observe these negative spill-overs within the report card scheme, particularly the graphic. Just eye-baling the graphs I couldn't see anything but I didn't have the data. My question is did you analyze that separately?

MS. KAPSAK: What we can do is we can look to see if there were significant differences within a particular scheme. I would have to go back and look myself but I think it is a great question.

DR. MURPHY: Thank you.

DR. SIMONE: I just have a quick question--

MS. FRASER: Excuse me--

DR. SIMONE: Dr. Simone.

MS. FRASER: Thank you.

DR. SIMONE: You did a great job. I want to congratulate you as well.

MS. KAPSAK: Thank you.

DR. SIMONE: You had a lot of stress doing this. Right?

MS. KAPSAK: Oh, yes.

DR. SIMONE: So, on a scale of 0-3, how much stress did you have?

[Laughter]

MS. KAPSAK: I think both of my bosses are sitting right here. Perhaps you could talk to them after this and I could get some kind of reward, or something.

[Laughter]

DR. SIMONE: You will get a reward, but give me a number on a scale of 0-3, how much stress?

MS. KAPSAK: How is my stress right now? It is zero because I am done.

[Laughter]

DR. SIMONE: Before you came here?

MS. KAPSAK: Before I came? We have given this presentation before so it is fine.

DR. SIMONE: The point I am trying to make is I think a lot of this information is very, very complicated.

MS. KAPSAK: Sure.

DR. SIMONE: I deal with cancer all the time and it is a very complicated subject and I try to simplify it all the time. So, if we can get some kind of a scale that is meaningful to people, like if I asked you the same questions, 0-3, you would probably tell me a 3 from the prior statement or two that you made. That is easily understandable by people. I think we need something easily understood by people--

MS. FRASER: Excuse me. I know you are scheduled to speak this afternoon, if you would save those comments for then. These are really clarifying questions at this opportunity. I would appreciate it if we would stay on that side, and welcome your comments at the opportunity this afternoon.

MS. KAPSAK: And certainly there could be more research about the number of levels which may help consumers, and how that is communicated.

MS. FRASER: Well, thank you very much for that wonderful presentation.

[Applause]

If you all are in agreement, I think we could do the next presentation before lunch, which will save us some more time after lunch for comments. So, with that, as we switch over, it is my pleasure to welcome from the lovely State of West Virginia and West Virginia University Dr. Paula Fitzgerald Bone and Dr. Karen Russo France to do an overview of their research findings. We thank them very much for traveling here to join us today.

West Virginia University Research

[see presentation slides for Drs. Bone and France]

DR. BONE: You have no idea how exciting it is for an academic to be here with everybody. We are thrilled to be here. My colleague, Karen France, will do the final part of the presentation and I will get you started.

We thought we were starting after lunch so I was going to review what it was that we were talking about--

[Laughter]

Obviously, we are dealing with Dietary Supplement Health Education Act. We are dealing with outcome of Pearson v. Shalala, and really looking at how do we best communicate information in the marketplace so consumers can make good decisions.

So, Karen and I got started on this because there was a change in the communication in the information environment. So, we started thinking about what is going on? What is it that consumers use, and how do they use this information that is out in the marketplace? We looked at what we called two paradigms of information decision-making, the direct effects model which seems to be how the world is operating, at least from the courts and from DSHEA, and then I will provide you with information and we will then incorporate that information into your decision-making accordingly.

Or, perhaps another paradigm that is important and comes out through the consumer behavior literature, and that is there are filters in interpreting this information in light of other things that are going on in my life. Existing literature, clearly DSHEA and the Pearson decision left us in an environment where there is a claim, a structure/function claim and a disclaimer--the FDA has not evaluated this claim, so forth and so on. The Pearson decision left us with the health claim--the disease claim is what we were calling it during the time we wrote this paper, and then the disclaimer on the level of scientific certainty.