Food Additives Permitted for Direct Addition to Food for Human Consumption; Olestra

Federal RegisterAug 5, 2003

Ask Donna

What actually matters in this document.

Text

DEPARTMENT OF HEALTH AND HUMAN SERVICES

Food and Drug Administration

21 CFR Part 172

[Docket No. 2000F-0792]

Food Additives Permitted for Direct Addition to Food for Human Consumption; Olestra

AGENCY:

Food and Drug Administration, HHS.

ACTION:

Final rule.

SUMMARY:

The Food and Drug Administration (FDA) is amending the food additive regulations to remove the requirement for the label statement prescribed specifically for savory snack products that contain olestra. This action is in response to a petition filed by the Procter and Gamble Co.

DATES:

The regulation is effective August 5, 2003. Submit written objections and requests for a hearing by September 4, 2003.

ADDRESSES:

Submit written objections to the Division of Dockets Management (HFA-305), Food and Drug Administration, rm. 1061, 5630 Fishers Lane, Rockville, MD 20852. Submit electronic objections to

http://www.fda.gov/dockets/ecomments.

FOR FURTHER INFORMATION CONTACT:

Mary D. Ditto, Center for Food Safety and Applied Nutrition (HFS-255), Food and Drug Administration, 5100 Paint Branch Pkwy., College Park, MD 20740-3835, 202-418-3102.

SUPPLEMENTARY INFORMATION:

Table of Contents

I. Subject of Petition

II. Background

A. Basis for Requiring the Label Statement—1996 Decision

1. Legal Authority for the 1996 Label Statement

2. GI Issues Associated With Olestra

3. Nutritional Issues Associated With Olestra

B. Opportunity for Comment and Consideration of New Data

1. Request for Comments on Label Statement Required by the 1996 Final Rule

2. P&G's Commitment To Further Studies

3. FDA's Commitment to Convene an FAC Meeting

4. P&G's Petition To Remove the Requirement for the Label Statement

5. Comments Received

III. Data and Information Since the 1996 Final Rule

A. Introduction

B. Surveys and Postmarket Passive Surveillance Regarding GI Effects

1. Telephone Surveys Regarding GI Complaints

a. P&G

b. CSPI

2. Postmarket Passive Surveillance by P&G

3. Postmarket Surveillance Reports From CSPI

4. Comments Regarding Consumer Reports

C. Studies Regarding GI Effects

1. Rechallenge Study

2. Acute Consumption Study

3. Home Consumption Study

4. Stool Composition Study

5. Comments Regarding the GI Studies

D. A Study Regarding Nutritional Effects—Active Surveillance

1. Active Surveillance Study by P&G

2. Comments Regarding the Active Surveillance Study

E. Consultations and Literature Review Regarding Nutritional Effects

F. Consumer Perception Studies of the Label Statement

1. 1996 Consumer Studies

2. 1999 Consumer Studies

G. 1998 FAC Discussion of the Label Statement

IV. FDA's Conclusions

A. The Applicable Legal Standard

B. FDA's Conclusions Regarding Gastrointestinal Effects

1. Basis of the 1996 Final Rule—GI Effects

a. Abdominal cramping

b. Loose stools

2. Data in the Current Petition—GI Effects

a. Abdominal cramping

b. Loose stools

C. FDA's Conclusions Regarding Nutritional Effects

1. Basis of the 1996 Final Rule—Nutritional Effects

2. Data in the Current Petition—Nutritional Effects

V. Response to Comments on the Label

A. Label Statement for GI Effects

B. Label Statement for Nutritional Effects

C. Labeling for Special Populations

D. Label Statement in Its Entirety

E. Data and Information Considered in this Rulemaking

F. Safety of Olestra

G. Allergenicity of Olestra or Olestra-Containing Foods

H. Nutrition Labeling and Claims

I. Appearance of the Label Statement

J. Labeling for Single-Serving Packages

K. 1995 and 1998 FAC Meetings

VI. Summary

VII. Environmental Impact

VIII. Inspection of Documents

IX. Objections

X. References

I. Subject of Petition

In a notice in the

Federal Register

of March 3, 2000 (65 FR 11585-11586), FDA announced that a food additive petition had been filed by the Procter & Gamble Co., 6071 Center Hill Ave., Cincinnati, OH 45224 (P&G, the petitioner) proposing that the food additive regulations be amended in § 172.867

Olestra

(21 CFR 172.867) to remove the requirement for the label statement prescribed in § 172.867(e).

II. Background

In the

Federal Register

of January 30, 1996 (61 FR 3118, “the 1996 final rule”) FDA announced the approval of olestra for use as a fat substitute in prepackaged ready-to-eat savory snacks. Olestra is the common name for a mixture of substances formed by chemical combination of sucrose with six, seven, or eight fatty acids. The fatty acids, bound to sucrose by ester bonds, are derived from edible fats and oils.

Olestra is essentially not absorbed or metabolized and passes unchanged through the gastrointestinal (GI) system (61 FR 3118 at 3125-3127). Therefore, olestra has the potential to affect GI physiology and function. Additionally, because of olestra's physical properties, fat-soluble nutrients present in olestra-containing foods

or other foods in the GI tract at the same time as olestra can partition into olestra and pass through the GI tract without being absorbed by the body. Therefore, FDA required the addition of fat-soluble vitamins A, D, E, and K, to savory snacks containing olestra to compensate for any inhibition of absorption by olestra (§ 172.867(d)).

Olestra has only been approved for use as a fat substitute in savory snacks. Throughout this document, we refer to olestra-containing foods to include those savory snacks made with olestra as well as other olestra-containing foods used in the preapproval studies for olestra.

At the time of the 1996 final rule, FDA concluded that, to avoid being misbranded within the meaning of sections 201(n) and 403(a)(1) of the Federal Food, Drug, and Cosmetic Act (the act) (21 U.S.C. 321(n) and 343(a)(1)), olestra-containing foods would need to bear a label statement to inform consumers about possible effects of olestra on the GI system. The label statement also would clarify that the added vitamins were present to compensate for any nutritional effects of olestra, rather than to provide enhanced nutritional value. Therefore, the 1996 final rule required that foods containing olestra be labeled with the following statement in a boxed format: “THIS PRODUCT CONTAINS OLESTRA. Olestra may cause abdominal cramping and loose stools. Olestra inhibits the absorption of some vitamins and other nutrients. Vitamins A, D, E, and K have been added.” (§ 172.867(e)(1)). FDA included the term “other nutrients” because any nutrient that is as lipophilic as these vitamins would also be affected, although there was no known basis for adding such nutrients back. The agency also required that the statement be made in a standardized format that specifies, among other things, type style and type size, and that

the label statement be surrounded by a box to ensure proper prominence. This requirement was established under section 409(c)(3)(B) of the act (21 U.S.C. 348(c)(3)(B)), which prohibits approval of a food additive if the proposed use would result in misbranding of food (61 FR 3118 at 3160). The legal authority and scientific basis that underlaid the requirement for this label statement are reviewed in detail in the next section of this document.

A. Basis for Requiring the Label Statement—1996 Decision

1. Legal Authority for the 1996 Label Statement

Under section 403(a)(1) of the act, a food is deemed to be misbranded if its labeling is false or misleading in any particular. Section 201(n) of the act amplifies what is meant by “misleading.” Section 201(n) of the act states that in determining whether labeling is misleading, the agency shall take into account not only representations made or suggested about the product, but also the extent to which the labeling fails to reveal facts material in light of such representations or material with respect to consequences which may result from use under the conditions of use prescribed in the labeling or under such conditions of use as are customary or usual (

see

21 CFR 1.21). Thus, the omission of such material fact from the label or labeling of a food causes the product to be misbranded within the meaning of sections 201(n) and 403(a)(1) of the act.

2. GI Issues Associated With Olestra

As noted, olestra is not digested or absorbed, and it passes through the GI tract intact. The petitioner conducted a number of studies to address issues of potential concern with respect to the effect of olestra as it passes through the GI tract (61 FR 3118 at 3152-3159). For example, during studies designed primarily to assess potential effects of olestra on absorption of fat-soluble dietary components present in the gut at the same time, the petitioner also assessed the potential for olestra to elicit GI symptoms such as cramping, bloating, loose stools, and diarrhea-like symptoms by collecting reports from participants in the studies. In two human nutritional studies

(88 and 100 subjects respectively), the entire diet of the subjects was controlled during the length of an 8-week study period. The studies were parallel, double-blind, and placebo-controlled, with olestra dosages of 0 (placebo), 8, 20, and 32 grams per day (g/d).

The diets were formulated so that the total digestible fat (triglyceride) content was the same for all treatment groups. Triglyceride was added into the diets in the form of butter, margarine, or vegetable oil to compensate for the amount of fat replaced by olestra in the olestra-containing foods. Olestra was added to various food items by substituting olestra for triglyceride in recipes or in cooking oils. Therefore, the total amount of lipid-like material (digestible triglyceride plus olestra) increased with increasing olestra dose. Each meal contained olestra or the corresponding placebo (triglyceride). Subjects were questioned daily about changes in their health, including GI symptoms. To facilitate collection of GI symptom data, a questionnaire provided a list of common GI symptoms along with general definitions of each and was completed by each subject to capture data about the type, severity, and duration of symptoms experienced. As noted in the 1996 final rule (61 FR 3118 at 3152), the petitioner stated that the two 8-week studies were not intended to examine GI symptoms under real-life consumption conditions where snacks are not consumed every day with every meal and where people may moderate intake if they experience GI symptoms.

In evaluating olestra's nutritional effects, the petitioner conducted two 8-week clinical studies, the 8-week clinical dose response study (8-week DR), and the 8-week clinical vitamin restoration study (8-week VR)(61 FR 3118 at 3133-3134). In this document, when discussing the combined results of these studies, they will be called the two 8-week studies.

By comparison, FDA concluded that the estimated lifetime-averaged daily intake at the 90th percentile of olestra consumption would be 7.0 grams per person per day (g/p/d) (61 FR 3118 at 3124).

FDA's analysis of the data from the two 8-week studies (61 FR 3118 at 3152-3154) showed that there was a dose-response effect for olestra with respect to two endpoints, reported diarrhea/loose stools and fecal urgency. Reporting of diarrhea was based on subjects' perception of diarrhea. FDA found no evidence that study subjects experiencing olestra-related symptoms described as “diarrhea” also experienced significant fluid or significant electrolyte loss. The effect of olestra on stool consistency is similar to that produced by liquid petrolatum, which softens fecal contents. FDA recognized that the effect observed was not diarrhea in the clinical sense, but used that term in the 1996 final rule, and is using that term here, because it is the term used in the study report. FDA also found that these GI symptoms cease soon after olestra is no longer consumed.

The petitioner also conducted a study, the Fecal Parameters Study, designed to examine fecal composition of stools from subjects who reported diarrhea when consuming olestra (61 FR 3118 at 3155). The study consisted of two phases, a screening phase and a study phase. The screening phase was conducted to identify subjects who reported GI symptoms from olestra consumption. During the study phase, the identified subjects ate different amounts of olestra, and GI symptoms were recorded and fecal measurements were made. From the initial screening phase, eighteen subjects reported an increase in the frequency, severity, or duration of GI symptoms during the olestra period, relative to the placebo period. These 18 subjects were selected to take part in the study phase, and 15 completed the study. The study phase was a crossover, placebo-controlled, single-blind (subject) design with three treatment groups, 0, 10, and 20 g/d olestra. Each subject received each treatment for 7 days. The treatment periods were separated by 7-day washout periods. Subjects ate all treatment meals under supervision at the clinical site, and ate their habitual diets at home during the washout periods. Study subjects recorded GI symptoms daily. Total fecal collections were made the last 3 days of each treatment period. Daily stool collections were measured for wet weight, volume, and density, and the pooled three day samples were analyzed for water concentration, dry weight, olestra content, sodium (Na), potassium (K), chloride (Cl), total and individual bile salts, free fatty acids, triglycerides, and total lipids.

Measurements of the concentration of stool water and electrolytes (Na, K, and Cl) suggested that these parameters did not differ in the stools of persons reporting “diarrhea” during the olestra 20 g/d period from those in the nondiarrheal stools (during the placebo period) of the same persons. However, it was not possible to analyze stool electrolyte values by individual stools or by individual days because the stools were pooled from the 3-day collection period, as is normally done when measuring fecal parameters. FDA noted that there appeared to be an increased weight of stools in those subjects reporting “diarrhea” when eating 20 g/d olestra that is not completely accounted for by the presence of olestra in the stools. FDA concluded that the results of this study indicated that there is no difference in stool composition (

e.g.

, water and electrolyte content) when subjects consumed olestra versus placebo (61 FR 3118 at 3155).

FDA found that the number of subjects in the Fecal Parameters Study

who reported diarrhea increased with increasing dose of olestra (

i.e.

, 3 subjects (20 percent) in the placebo, 6 subjects (40 percent) who consumed 10 g of olestra, and 11 subjects (69 percent) who consumed 20 g of olestra). In addition, both the mean number of reported diarrheal bowel movements per subject reporting any diarrhea, and the severity of the reported diarrhea, increased with increasing olestra consumption. Although there was an increase in the number of subjects reporting loose stools with increasing olestra dose, this increase was not statistically significant. FDA concluded that these results were qualitatively similar to the results of the 8-week studies.

The agency concluded, based upon its evaluation of the data and information available at the time, that consumption of olestra causes GI effects such as loose stools, abdominal cramping, and diarrhea-like symptoms. Additionally, the agency concluded that while olestra caused these GI symptoms, there was no evidence that these effects represented adverse health consequences.

At the time of approval, the agency did not have information about the potential GI effects from usual or customary consumption of olestra in savory snacks. Nonetheless, FDA considered it prudent to rely on the available data in deciding whether a label statement about olestra's potential effects on the GI tract was necessary. Olestra had the potential to be consumed in relatively large quantities by every segment of the U.S. population. Additionally, because olestra had never before been available in the marketplace, consumers had no experience with it and were not familiar with it or its potential to cause GI effects. The agency believed that providing consumers label information about olestra's GI effects would preclude unnecessary concerns about the origin of GI effects, were they to be observed, and might also prevent unnecessary or inappropriate medical treatment of those symptoms (61 FR 3118 at 3161). Based on the weight of the evidence about olestra's potential to cause GI effects, as well as the agency's belief that consumers lacked familiarity with olestra and its potential to cause such effects, FDA concluded at the time of olestra's approval that the relationship between GI symptoms and consumption of foods containing olestra is a fact that is material in light of the consequences of consuming olestra, and therefore a label statement was required.

3. Nutritional Issues Associated With Olestra

FDA concluded that olestra inhibits the absorption of the fat-soluble components of the diet when these components are present in the GI tract simultaneously with olestra (61 FR 3118 at 3132-3147). Such components include the fat-soluble vitamins A, D, E, and K, and the lipophilic carotenoids. Based on the data from the nutritional studies, FDA concluded that addition of the four fat-soluble vitamins (A, D, E, K) to foods containing olestra would compensate for any decreased absorption due to the action of olestra, thus ensuring that consumption of an olestra-containing food would not alter the amount of vitamin available for absorption (61 FR 3118 at 3144-3147). The amounts of the vitamins to be provided are prescribed to ensure safe use (§ 172.867(d)). As required under section 403(i) of the act, these vitamins are declared in the ingredient listing.

The added vitamins were not to be considered in determining nutrient content of the food for the nutritional label or for any nutrient claims, expressed or implied. This is because the added vitamins simply compensate for the transient impaired absorption of vitamins A, D, E, and K,

i.e.

, they are added to ensure no change (neither increase nor decrease) in vitamin availability. Thus, the vitamins added to olestra do not contribute significant amounts of these nutrients to the diet (61 FR 3118 at 3161).

Labeling may be considered misleading not only if it fails to reveal facts that are material in light of consequences that may result from use of a food, but also if the labeling fails to reveal facts that are material in light of representations made. Therefore, to set the context for why vitamins A, D, E, and K were added, FDA required a label statement providing information both that vitamins A, D, E, and K had been added and that olestra inhibits the absorption of vitamins. Because FDA believed that consumers who see vitamins A, D, E, and K in the ingredient listing might incorrectly believe that the food was fortified with these vitamins, the agency required an explanatory statement on the label of olestra-containing foods to inform consumers that olestra-containing foods were not an enhanced source of vitamins A, D, E, and K. The statement indicated that olestra inhibits the absorption of vitamins and other nutrients to explain why they were added. FDA included the term “other nutrients” because any nutrient that is as lipophilic as these vitamins would also be affected, although FDA concluded that there was no basis for adding back nutrients other than vitamins A, D, E, and K. In this way, FDA sought to make clear to consumers the reason for the presence of vitamins A, D, E, and K in the ingredient listing.

Carotenoids are fat-soluble components in the diet, the majority of which are derived from fruits and vegetables. Data from the petitioner's two 8-week studies demonstrated that consumption of olestra inhibits absorption of carotenoids as measured by a decrease in serum carotenoid levels (61 FR 3118 at 3147-3149). Co-consumption of olestra and a carotenoid-containing food allows the greatest interaction between olestra and the carotenoid, thereby maximizing the potential for interfering with absorption of the carotenoid from the GI tract.

Beta-carotene is a provitamin A carotenoid that is a dietary source of vitamin A; provitamin A carotenoids are converted in the body into vitamin A. At the time of the 1996 final rule, FDA concluded that supplementing olestra-containing foods with vitamin A would compensate for olestra's effects on the provitamin A function of carotenoids.

In evaluating whether there is a scientific basis to require the addition of any carotenoids to olestra-containing foods, FDA consulted with scientists at the National Cancer Institute of the National Institutes of Health (NIH) and the National Eye Institute (NEI) of the NIH (61 FR 3118 at 3148-3149), and the agency's Food Advisory Committee (FAC) (61 FR 3118 at 3121). At the 1995 FAC meeting on olestra, experts with a range of views discussed whether carotenoids themselves have beneficial health effects, or whether it is some other substance in fruits and vegetables that provides the claimed health effects, in which case the carotenoids are serving solely as markers for fruit and vegetable consumption. Five different conferences or reviewing groups preceding the 1995 FAC meeting had examined the relationship between carotenoids and disease. All of these groups had concluded that there was insufficient evidence to recommend consumption of carotenoids, except to encourage the consumption of fruits and vegetables (61 FR 3118 at 3148). Although epidemiological studies showed an association between diets rich in fruits and vegetables (including those that contain carotenoids) and decreased cancer risk, there was no direct evidence that carotenoids themselves were responsible for or contributed in a significant way to that protective benefit. Therefore, at the time of the approval of olestra, the agency

concluded that the available data did not establish any identifiable nutritional or prophylactic benefits for carotenoids, either individually or collectively, aside from the provitamin A function (61 FR 3118 at 3147-3149).

Thus, FDA found no scientific basis for requiring the addition of any carotenoid to olestra-containing foods. The agency also found that the actual magnitude of olestra's effects on carotenoid absorption was likely to be within the range of the normal variation of such absorption due to diet and bioavailability, providing additional assurance that the effect of olestra on the absorption of carotenoids did not raise concern. Accordingly, FDA concluded that there was no basis for requiring a statement about carotenoids on the label of olestra-containing food.

B. Opportunity for Comment and Consideration of New Data

1. Request for Comments on the Label Statement Required by the 1996 Final Rule

Because section 409 of the act prohibits, among other things, approval of a food additive if doing so would cause misbranding, the agency concluded that the olestra label statement should be imposed as a requirement as part of the food additive petition process (§ 172.867(e)). The agency acknowledged, however, that the specific wording had not been tested or subject to an opportunity for comment. Thus, the agency requested comments on the label statement from interested persons on such issues as the need for labeling, the adequacy of its content, and the agency's current word choices (61 FR 3118 at 3160).

After the publication of the 1996 final rule, the agency received timely comments on the label statement,

as well as objections to the 1996 final rule.

The comments received by April 1, 1996, included the results of P&G's consumer focus group studies on the label statement. Frito-Lay also submitted consumer perception studies on the olestra label statement. These studies are discussed in section III.F.1 of this document. The agency continued to receive comments on the label statement after April 1, 1996. In this document, FDA addresses comments received on the label statement regardless of whether the comments were received by April 1, 1996.

Timely objections were to be filed by February 29, 1996. FDA's response to these objections and requests for hearing is published elsewhere in this issue of the

Federal Register.

2. P&G's Commitment To Further Studies

In a letter to the agency dated January 24, 1996, the petitioner stated its intention to conduct focus group testing of the required olestra label statement, to establish a postmarket surveillance system, to conduct additional studies of olestra exposure (both amounts consumed and patterns of consumption), and to conduct additional studies regarding the effects of olestra consumption (61 FR 3118 at 3160 and 3168). FDA responded that P&G was to conduct the studies it had identified in its letter to FDA, consistent with the timetables identified in that letter (61 FR 3118 at 3168).

P&G did carry out the surveillance and studies outlined in its letter of commitment, and performed additional studies not mentioned in the January 1996 letter. After the publication of the 1996 final rule, P&G carried out its commitment to establish a system of passive surveillance to collect spontaneous reports of possible effects that consumers associated with the consumption of olestra-containing snacks. This system included establishing an outside panel of medical experts to review reports, followup on reports of serious illness, and provide FDA information about reports received.

P&G also carried out its commitment to conduct studies on the exposure and effects of olestra. The active surveillance program that P&G sponsored was designed to examine the impact of olestra consumption on endpoints such as serum concentrations of carotenoids and vitamins, olestra consumption patterns (including frequency and amounts), and GI symptoms.

These data and information were presented to the FAC in June 1998, and were eventually incorporated into the petition that is the subject of this rulemaking.

3. FDA's Commitment To Convene an FAC Meeting

In the 1996 final rule, FDA committed to review and evaluate any new data and information bearing on the safety of olestra and to present such information to the agency's FAC within 30 months of the approval of the use of olestra in savory snacks (61 FR 3118 at 3168-3169; § 172.867(f)).

FDA convened a meeting of its FAC within 30 months of the approval of the use of olestra in savory snacks. At an open public meeting, held June 15-17, 1998, new data and information concerning olestra, obtained since the 1996 approval were presented (Ref. 1). These new data, which comprise the majority of material that P&G subsequently submitted in its petition, are discussed in section III of this document. FDA, P&G, the Center for Science in the Public Interest (CSPI), and other interested members of the public made presentations to the Committee. After presentation of the new data, the FAC discussed the label statement specified in § 172.867(e). The complete set of transcripts of the June 1998 FAC meeting (“the transcript” or “transcript”) is publicly available through FDA's Division of Dockets Management and through FDA's Internet site.

The Internet site is located at

http://www.fda.gov/ohrms/dockets/ac/cfsan98t.htm#Food Advisory Committee

(choose June 15, 16, and 17).

4. P&G's Petition To Remove the Requirement for the Label Statement

P&G submitted a food additive petition, dated December 1, 1999, to amend the food additive regulations in § 172.867

Olestra

by removing the requirement for the label statement prescribed in § 172.867(e). This petition incorporated the studies and information that were performed after the publication of the 1996 final rule. As noted, much of that material was discussed by the FAC in 1998.

5. Comments Received

FDA received approximately 80 letters, each containing one or more comments, on the olestra label statement.

Some of the comments were submitted in response to FDA's request in the 1996 final rule for comments on the olestra label statement. Other comments were submitted in response to the January 24, 1996, announcement of the approval of olestra for use in savory snacks. Because all of these comments addressed P&G's original petition, which was granted in 1996, in this document FDA refers to these comments as comments “to the 1996 final rule.” Comments were also submitted in response to publication in the

Federal Register

of the filing notice for the current petition (65 FR 11585, March 3, 2000); in this document, FDA refers to these comments as comments “to the current petition.”

FDA notes that one of the objections submitted by CSPI concerns the label statement required by the 1996 final rule.

Although not part of the petition being considered in the current rulemaking, FDA reviewed comments regarding labeling that were addressed to the 1998 FAC. These comments raised no substantive issue that was not already considered as part of the current food additive petition.

Comments were submitted by P&G, Frito-Lay, Inc. (Frito-Lay), and other members of the food industry, as well as from individual consumers, consumer organizations, academia, trade associations, and a member of Congress. Several comments were filed by CSPI. Several parties, including Frito-Lay and

CSPI, submitted comments to both the 1996 final rule and the current petition.

Although section 409 of the act establishes no comment period for food additive petitions, and the agency generally does not solicit comments in notices announcing the filing of a food additive petition, it is FDA's practice to consider any relevant comments submitted prior to the agency's decision on a petition.

In this document, FDA separately discusses data from telephone surveys and passive surveillance regarding GI effects, new studies regarding GI effects, and active surveillance and other information regarding nutritional effects of olestra. As part of its discussion of these areas, FDA describes and responds to comments relevant to the topic. FDA discusses and responds to comments on other topics (such as the wording of the label statement, the prominence and placement of the label statement, and the need for a label statement) in a separate section (see section V of this document). In responding to the submitted comments, FDA has considered all of the data and information available in the record that bear on the olestra label statement, including the data and information in the 1996 final rule as well as the new information in the current petition.

See discussion of comments in the filing notice for this petition (65 FR 11585-11586, March 3, 2000).

III. Data and Information Since the 1996 Final Rule

A. Introduction

Olestra-containing snacks were introduced into test markets in April 1996, and national marketing began in February 1998. P&G established a system of passive surveillance to collect reports of possible effects that consumers associated with eating olestra-containing snacks. This surveillance system was in place when test marketing began. P&G has submitted reports to the agency, as well as analyses of such reports. P&G also established its program for active surveillance to monitor, among other things, possible nutritional impacts of olestra consumption.

P&G conducted studies concerning possible GI effects from consuming olestra-containing snacks in “real-life” situations. Specifically, P&G conducted four controlled studies concerning possible GI effects in humans and submitted reports about those studies to FDA.

The four controlled studies are described as follows:

A copy of the petition submitted by P&G, as well as copies of the studies, surveys, and other supporting materials listed here, can be found at the Division of Dockets Management, Docket No. 00F-0792.

• An Acute Consumption Study,

• A Six-Week Consumption Facilitated

Ad Lib

Study (also called the Home Consumption Study),

• A Rechallenge Study, and

• A Stool Composition Study.

In the current petition, P&G also submitted the following data and information:

• Reports and analyses of data collected through consumer focus group and perception studies,

• Surveys regarding GI symptoms,

• Updated literature reviews on carotenoids and disease, and

• A report and analysis of the first year of data collected in an Active Surveillance Study.

CSPI also submitted reports from individuals who attributed an effect to the consumption of an olestra-containing food (Docket No. 87F-0179). In some cases, CSPI obtained medical records from consumers and forwarded them to FDA for analysis.

Below, FDA describes in detail the studies and information submitted in support of this petition, comments to the 1996 final rule that discuss the labeling of olestra-containing foods, comments to the current petition, and other relevant information.

B. Surveys and Postmarket Passive Surveillance Regarding GI Effects

1. Telephone Surveys Regarding GI Complaints

P&G.

P&G sponsored two telephone surveys to investigate the frequency and severity of GI complaints, to investigate the frequency of consumption of foods that consumers believe cause GI symptoms, and to determine knowledge about reported GI symptoms from olestra-containing foods. Both of these surveys were performed before olestra-containing foods were available for sale in those markets. The first survey was done in February 1997 in Indiana (in Marion County, where Indianapolis is located), which was later a test-market for olestra-containing foods. This survey also served as a pilot study for the second survey, which was a national survey of the U.S. population completed in September 1997. National marketing of olestra-containing foods began in February 1998.

The petitioner acknowledged limitations in the design of the first survey completed in Indiana. For example, because the first survey was also the pilot study, the study instrument had not yet been validated.

Also, the sample size was small and not shown to be representative of the general population surveyed. Despite these limitations, the petitioner concluded from the survey that GI symptoms were very common among the adult respondents polled. Asked about the previous three month period, respondents reported most frequently the GI symptoms of gas (34.6 percent), diarrhea (33.2 percent), and abdominal cramps (25.8 percent). Of those respondents who experienced one or more GI symptoms, 14 percent reported seeking medical attention because of the symptom. More than half of those who experienced a GI symptom said it was of moderate to severe intensity. The other result noted by the petitioner in this study was that there are a number of common foods (

e.g.,

beans, onions, spicy foods) that respondents said caused them to have GI symptoms, but more than 80 percent of these respondents said they continued to eat these foods. Approximately half of the respondents had heard of olestra and among that group, 18 to 28 percent associated olestra with a GI symptom such as abdominal cramping or diarrhea.

Data obtained from the first survey was used to validate the study instrument for use in the second survey.

The second survey was a larger, national survey that was designed with a reliability check, and a portion of the survey was designed to be a truly random sample of respondents. In this survey, 40.5 percent of respondents reported having one or more GI symptoms in the previous month. Of those respondents reporting a GI symptom, 21.8 percent had abdominal pain or discomfort, and 26.9 percent reported diarrhea or loose stools. More than 65 percent of respondents rated each of their symptoms as moderate to severe in intensity, and 14 percent consulted physicians about their symptoms. When asked about specific foods, respondents reported having GI symptoms after eating foods such as beans (22 percent) or spicy foods (34.4 percent). Despite symptoms, approximately 80 percent continued to consume these foods. The petitioner also found that women were more likely than men to report GI symptoms, and that abdominal pain, discomfort, and bloating were more commonly reported by women. There was little difference between males and females for reports of diarrhea. More than half the respondents in this study had heard of olestra, and of those, a varying number associated olestra with different GI

symptoms (diarrhea (33 percent), loose stools (4 percent), cramps (11 percent), other GI problems (23 percent)).

FDA notes that in both these surveys, the percent of individuals reporting a GI symptom was high. FDA also notes that while the great majority of respondents in both surveys indicated that they consumed foods that caused them GI symptoms, the survey did not obtain information about the severity of these symptoms (Ref. 2).

CSPI.

In 1996, CSPI commissioned a telephone survey in cities where olestra-containing foods were test marketed (Cedar Rapids, IA, Eau Claire, WI, and Grand Junction, CO), and submitted a report of the results to FDA.

The purpose of CSPI's survey was to determine how many people in the test market had tried olestra-containing snacks, and of those who had eaten olestra-containing snacks, how many had experienced GI symptoms. A random digit dialing sampling system was used, until a total of 506 telephone interviews were conducted in June and July 1996. CSPI said that 27 percent of individuals surveyed had tried the newly marketed olestra-containing chips, and of those, 20 percent reported experiencing GI symptoms characterized by CSPI as an adverse GI effect. Respondents characterized these events as mild (58 percent), moderate (23 percent), or severe (9 percent). CSPI also found that the majority of respondents (78 percent) had seen negative reports in the press about olestra, although only 28 percent said they were concerned about these possible effects. Based on this telephone survey, CSPI predicted that a large number of adverse events would be caused by the national marketing of olestra-containing foods. CSPI performed a separate survey of the original respondents who had not eaten olestra-containing chips but ate other chips to assess the frequency of GI events associated with consumption of conventional savory snacks and found that only 0.5 percent associated an adverse GI effect with eating a triglyceride savory snack.

CSPI submitted this report to Docket No. 87F-0179.

CSPI commissioned a second survey, which was conducted in April and May 1997 in the Indianapolis area where olestra-containing foods were test-marketed. The purpose of the survey was to ascertain the consumption of olestra-containing foods and possible rate of adverse effects. CSPI submitted a report of the results to FDA.

CSPI reported that the majority of respondents said they had eaten savory snacks in the past 8 weeks (68.8 percent), and that a smaller portion of respondents (32.7 percent) said they had eaten olestra-containing foods. CSPI said that when respondents were asked whether they had eaten a specific brand of chips that contain olestra, they said yes, but when asked whether they had eaten olestra, these consumers responded that they had not eaten olestra. Of the group of respondents who had eaten an olestra-containing food, 8.3 percent reported experiencing what was characterized as an adverse effect after eating the olestra-containing food.

CSPI submitted this report to the Docket No. 87F-0179.

In its review of the data and information submitted by CSPI, FDA found that the studies provided information about the prevalence of use of olestra-containing foods, awareness of GI symptoms associated with olestra, and sources of information that consumers were using to learn about GI symptoms associated with olestra. FDA disagreed, however, that these studies could provide information about the cause of these GI symptoms. FDA found that the study design was inadequate to determine the cause of GI symptoms. Additionally, it is known from food-borne illness outbreaks that attribution biases can influence consumers and lead to the erroneous attribution of symptoms to a particular food. CSPI's survey does not allow for the evaluation of erroneous attribution of GI symptoms to consumption of olestra-containing chips,

i.e.,

other plausible causes for illness reports were not considered (Ref. 3).

2. Postmarket Passive Surveillance by P&G

P&G established a system of passive surveillance to collect consumer reports associated with the consumption of olestra-containing foods. Passive surveillance refers to the collecting of spontaneous, voluntary reports about a product. In its petition, P&G presents an overview of both the utility and limitations of data obtained from spontaneous, postmarket consumer reports. P&G characterized postmarket passive surveillance as a means of identifying and characterizing potential issues, including safety issues, once a product has entered the marketplace and been utilized by the population at large. The population experience with a product will be much broader than that derived during premarket testing because the number of individuals participating in premarket testing is necessarily limited. A reporting rate may be calculated based on the total amount of product sold and the number of reports received. P&G notes, however, that there are a number of limitations with passive surveillance reporting. For example, these data do not lend themselves to assessment of causality because of the lack of controlled conditions and various confounding factors, such as a high background incidence of reported GI effects. Voluntary reporting, such as that obtained in P&G's passive surveillance system, is also subject to a number of biases including the level of attention the subject is receiving in the news media. Because reporting is voluntary and subject to interpretation, and because the total number of “exposures” can only be estimated, a true incidence rate cannot be calculated from passive surveillance.

Reports to P&G under its passive surveillance program for olestra were, for the most part, collected via calls made by consumers to a toll-free telephone number displayed on olestra-containing foods. Such calls were taken directly by P&G via its own toll-free telephone number. Calls were also forwarded to P&G by other snack food manufacturers (specifically, Frito-Lay). In some cases, information from such calls (but not the calls themselves) was forwarded to P&G by other snack manufacturers. Information was collected from callers as to the product used, specific product code information (where available), amount consumed, the nature of the complaint, symptom onset and duration, recurrence, characteristics and treatment, concomitant medications, and physician or other health professional involvement. Where physician contact was involved, or a medically significant event was reported, the petitioner attempted to obtain more detailed information including release of medical records for evaluation.

Consumer reports were reviewed by trained medical affairs staff at P&G. Additionally, the petitioner established a committee (Olestra Postmarketing Surveillance Committee) of medical experts outside of P&G whose membership included specialists in GI disease (both adult and pediatric GI), epidemiology, and pharmacology to review the reports received, and to make recommendations about the implications, if any, of these reports on the safety of olestra. The petitioner submitted reports to FDA of complaints associated with olestra consumption beginning in April 1996 with the test marketing of snacks made with olestra.

The petitioner reported that there were peaks in the number of reports received after the test marketing and

national introduction of olestra-containing foods. The petitioner found that while the absolute number of reports increased when olestra-containing foods were first introduced into test markets, the reporting rate (reports per amount of product sold) declined over time. The petitioner also found that over time the absolute number of reports declined and eventually reached a plateau. The greatest number of reports the petitioner received over a four month period was 4,951 in 1998 at the national introduction of olestra-containing snacks. The petitioner stated in its initial reports on passive surveillance that it is common to receive calls and complaints for products. At the start of national marketing the reporting rate was one report for approximately 100,000 servings sold, and 2 years later the reporting rate was one report for approximately 1 million servings sold.

The petitioner analyzed the data from its passive surveillance efforts and found no trend toward increased symptoms with increased consumption; no trend towards increased severity with increased consumption; and no difference in severity by age group or gender. The petitioner's Olestra Postmarketing Surveillance Committee reviewed reports using an algorithm developed to assess the likelihood that an effect was caused by olestra. Using this algorithm, P&G's committee concluded that many reports were not likely to be related to olestra and no serious reports could be attributed to olestra (Docket No. 00F-0792, submission dated March 3, 2000). Based on the nature of the complaints received, the petitioner designed subsequent studies to address, in part, issues arising from consumers' anecdotal reports after eating olestra-containing foods.

In the 1996 final rule, FDA determined that the use of olestra was safe based on results from the preapproval safety studies. FDA stated in the 1996 final rule (61 FR 3118 at 3168) that P&G's plans to continue to study the consumption and effects of olestra were both prudent and responsible. FDA expected, based on results of the preapproval studies, that some reports concerning GI upset, such as loose stools and abdominal cramping, would be collected through a system of passive surveillance. FDA also considered that postmarket passive surveillance had the potential to detect low frequency and unexpected events because postmarket passive surveillance involves the entire population that consumes a product.

FDA reviewed the reports of effects attributed to olestra and found that the majority of reports received concern GI effects such as loose stools and abdominal cramping. Other symptoms were reported at lower frequencies.

FDA recognizes that passive surveillance data such as those collected by P&G have utility but are limited in that they do not allow for the determination of a causal association between the product consumed (

i.e.

, olestra-containing foods) and effects reported. Such reports can, however, lead to hypothesis generation about why specific effects are occurring. This may then result in the development of studies used to test these hypotheses (Ref. 4).

FDA reviewed reports of effects associated with ingestion of olestra that led consumers to seek medical attention. Where possible, the agency reviewed medical records that were obtained directly from individuals who reported the effect or that were obtained through P&G and CSPI (Refs. 5 and 6). At the 1998 FAC meeting, FDA presented its analysis of the medical reports received up to that time. Among the reports discussed was a case where a consumer had undergone an appendectomy and associated this with consuming an olestra-containing food. The pathology diagnosis at the hospital noted that there were minimal inflammatory changes. FDA obtained a medical release from the patient in order to examine the pathology slides from the appendectomy, which were read independently by four FDA pathologists, all of whom confirmed the presence of inflammatory cells throughout the wall of the appendix meriting a diagnosis of acute appendicitis.

In many other of the medical records reviewed, physicians attributed patient symptoms to an etiology other than olestra, or did not provide an etiology. There were cases where physicians did attribute symptoms to olestra, but the limitations of passive surveillance make it difficult, if not impossible, to draw definitive conclusions about causality based on the review of individual medical records.

Transcript, vol. 1, pp. 271-276.

3. Postmarket Surveillance Reports From CSPI

CSPI has periodically submitted to FDA reports of effects allegedly associated with the consumption of olestra (Docket No. 87F-0179). The complaints were gathered initially in test markets by calls to an advertised toll-free telephone line, and gathered subsequently through CSPI's Internet site.

FDA analyzed the reports from CSPI and compared the information and analysis to the information and analysis of the reports submitted to FDA by P&G. FDA noted that the reports received by CSPI were very similar in nature, type of complaint, and amount consumed as the reports by P&G.

As discussed previously, passive surveillance has limited utility in determining causality.

Transcript, vol. 1, pp. 258-270.

4. Comments Regarding Consumer Reports

FDA received comments about reports of effects that consumers attributed to olestra. FDA considered these comments and responds in the following section of this document.

(Comment 1) One comment from an individual consumer to the current petition reported that a family member who had Addison's disease suffered gastric cramps and diarrhea, laid down, went into an Addisonian crisis, and died after consuming olestra-containing potato chips. The comment did not provide any further information regarding the death mentioned. CSPI forwarded to the agency the medical record of an Addison's disease patient who had reportedly consumed olestra-containing potato chips prior to death. The patient who died was diagnosed as having Addison's disease (adrenocortical insufficiency) and hypothyroidism in 1989. FDA believes, based on several factors, that the comment and medical record provided by CSPI refer to the same person.

FDA reviewed the medical record forwarded by CSPI. This patient collapsed suddenly after experiencing a bout of gastroenteritis, and reportedly consumed olestra-containing chips prior to the bout of gastroenteritis. An autopsy showed that the adrenal glands could not be identified and noted a finding of Hashimoto's thyroiditis. The medical record did not provide any information regarding the gastroenteritis experienced prior to death.

Due to the lack of information contained in the medical record, the agency was unable to determine whether the ingestion of olestra had any role in this patient's illness or death (Ref. 5).

The Certificate of Death lists acute cardiorespiratory arrest as the immediate cause of death with autoimmune adrenocortical deficiency syndrome as the underlying cause leading to the immediate cause of death. Other significant conditions contributing to death but not resulting in the underlying cause include mitral valve prolapse and Hashimoto's thyroiditis.

FDA investigated the death mentioned in the comment. As part of the investigation, FDA spoke with the patient's spouse and the patient's co-

workers about the events leading up to the patient's death.

(Comment 2) A comment from CSPI to the current petition asserted that FDA has never conducted indepth investigations (beyond reviewing the medical records of a few individuals) of any of the anecdotal reports, including those reports involving rectal bleeding, hospitalization, and death. The comment further asserted that the agency has ignored all of the anecdotal reports and has said that there is no proof that any of the reports were due to olestra. The comment also stated that in some reports, a patient's physician attributed his/her symptoms to olestra. The comment also quoted FDA review memoranda (Refs. 6 and 7) stating that olestra may have been responsible for some of the effects reported.

FDA does not agree that it has ignored anecdotal reports. Nor does it agree that it must conduct further investigation of the anecdotal reports. FDA regularly reviews the reports forwarded to the agency by P&G and CSPI. The reports are analyzed and summarized using criteria such as sex, age, symptoms reported, duration of symptoms, and amount of olestra-containing food consumed. The agency also reviews any medical records forwarded with the reports. CSPI provided no evidence to support its allegation that anecdotal reports have been ignored. Nor has CSPI provided any reason to suspect that serious adverse health effects could have been caused by olestra. The agency will continue to monitor reports as they are forwarded to the agency.

As stated in the memoranda cited by the comment, some of the effects reported may have been caused by olestra. These reports were collected using passive surveillance. As noted previously, passive surveillance is useful in that it can lead to hypothesis generation about why specific effects are occurring. These hypotheses can then be tested in controlled clinical trials. For example, the reports received from consumers attributing their symptoms to olestra served as a basis for some of the hypotheses tested in the petitioner's most recent controlled clinical studies. However, while passive surveillance data have utility, such data are limited in that they do not allow for the determination of a causal association between the product consumed and effects reported. Thus, based on the passive surveillance data alone, it is not possible to determine whether the effects reported were caused by consumption of an olestra-containing food.

The agency has reviewed all of the medical records that it has received from CSPI and P&G about consumers who saw a physician for an effect attributed to the consumption of an olestra-containing food. In fact, FDA has conducted an investigation into the death mentioned in the previous comment. FDA has also obtained and examined the pathology slides of a patient who had undergone an appendectomy that the patient associated with consumption of an olestra-containing food. The agency has not found sufficient evidence to conclude that olestra is likely to have caused the symptoms that led the consumers to see a physician.

Ref. 5 and Transcript, vol. 1, pp. 271-276.

(Comment 3) A comment from CSPI to the current petition stated that letters and electronic mail messages sent to P&G describing GI symptoms have not been included in reports that P&G submitted to the agency; therefore, the agency has not received all of the symptom-related reports. The comment recommended that the agency investigate whether it had received all reports.

FDA recognizes that not every report from a consumer will provide enough information for FDA to determine whether an effect was possibly related to olestra and that some judgement is needed in compiling data. In light of this limitation, the agency recognizes that P&G may not forward all reports it receives, such as those reports containing incomplete information, to the agency. While this may mean that less than 100 percent of reports are collected, the agency has no reason to believe that the complaints not forwarded to the agency constitute a unique data set or raise an issue not previously considered. Indeed, the reports gathered and forwarded independently to the agency by P&G and CSPI are consistent in terms of the nature of the complaints and the amounts of olestra consumed. CSPI's comment provides no specific information that would lead the agency to conclude that it has not received an accurate and representative sample of the effects reported to P&G or that such reports raise an issue not already considered. Thus, the agency finds that there is no basis for concluding that it should obtain and evaluate each and every report that P&G receives.

(Comment 4) A comment from CSPI to the current petition stated that the agency should obtain and disclose to the public the number of consumers (without identifying any particular individuals) who attributed their symptoms to olestra and reached an out-of-court settlement with the petitioner. The comment also asked for the number of consumers who, after attributing their symptoms to olestra, received or were offered reimbursement for their medical expenses. The comment requested that the agency consider this information in its rulemaking.

FDA does not agree that it should obtain and disclose to the public the number of consumers who attributed their symptoms to olestra and reached an out-of-court settlement with the petitioner. Importantly, the comment does not demonstrate the relevance of the requested information to the question at issue:

i.e.

, whether FDA should continue to require special labeling for olestra-containing foods. Moreover, settlement of lawsuits may be reached for a variety of reasons, including improved public relations or avoidance of unnecessary conflict, and do not address any factual issues regarding whether olestra is capable of causing the effects claimed.

C. Studies Regarding GI Effects

1. Rechallenge Study

The petitioner submitted a report of a study designed to test whether individuals who complained of a GI effect after consuming olestra-containing snacks would have the same experience with subsequent exposure (Refs. 8 and 9). This test was designed to show whether the GI effect is consistently associated with consumption of the olestra-containing snack. The petitioner's study was a randomized, double-blind, placebo-controlled, four-period, within-subject crossover study. Subjects were recruited from consumers who had voluntarily called the snack manufacturer and reported GI symptoms associated with consumption of olestra-containing snacks. Each subject made four visits to the study site, at least 1 week apart, and was provided with 2 ounces (oz) of either potato chips containing olestra (olestra chips) or potato chips containing conventional triglyceride

(triglyceride chips). At each visit, subjects were to consume as much as they could of the 2 oz serving. Each participant was randomly assigned to receive olestra chips at two visits and triglyceride chips at two visits. Participants were contacted after each visit and asked whether they had

experienced any GI symptoms within the week after eating the potato chip product.

In the various reports submitted by the petitioner, the terms triglyceride, full-fat, regular, and conventional were all used to describe the oil used in savory snacks. These terms mean the same thing. For consistency in this document, we use the word triglyceride.

The study was completed with 98 participants, the majority of whom had initially called to report that they had experienced diarrhea, loose stools, and/or abdominal cramping (61 percent, 16 percent, and 64 percent respectively). Approximately 48 percent of the participants described the symptoms that prompted their original call as severe. For nearly three-quarters of the participants, the amount of olestra chips consumed in the study was comparable to, or greater than, the amount associated with their initial call.

The petitioner found that during the study there were no significant differences following consumption of olestra chips, compared with consumption of triglyceride chips, in the frequencies of abdominal cramping (12 percent with olestra, 9 percent with triglyceride), diarrhea or loose stools (11 percent with olestra, 15 percent with triglyceride), gas (7 percent with olestra, 5 percent with triglyceride), or any other GI symptom (28 percent with olestra, 26 percent with triglyceride). Overall symptom severity ratings for all subjects were similar after consumption of olestra and triglyceride chips. The petitioner concluded that this study provided evidence that an episode that was initially reported to be an olestra-related effect was in all likelihood not olestra-related, and that there was no evidence of a population or subpopulation with a sensitivity to olestra. The petitioner suggested that these results indicate that initial calls made to the toll-free telephone line may reflect false attribution of symptoms to products made with olestra.

FDA found this study was adequately representative of the population who called the postmarketing surveillance system in terms of severity of initial symptoms and amount of olestra reportedly consumed prior to the initial symptom episode. FDA noted that while 98 participants were enrolled in the study, only 92 completed all 4 visits. The six dropouts were unrelated to olestra-related effects. Based on an analysis of the data in the study, FDA concluded that under the conditions of the study (two exposures of up to 2 oz of olestra-containing chips separated by at least a week), subjects eating olestra-containing chips were no more likely to report having had loose stools, abdominal cramps, or any other GI symptom compared to subjects eating an equivalent amount of triglyceride chips (Refs. 10 and 11).

2. Acute Consumption Study

P&G sponsored

a study to determine whether there was a difference in the nature or frequency of GI symptoms experienced by subjects eating olestra chips compared to those eating triglyceride chips, ad libitum on a single eating occasion (Ref. 12). The study was a randomized, placebo-controlled, double-blind study in which 1,092 adults and teenagers who were provided with a 13 oz bag of potato chips (either olestra chips or triglyceride chips) in a plain, unlabeled white bag, consumed as many chips as they desired while viewing a movie. Participants were also provided a 32-oz soft drink of their choice. Participants were told prior to the test that they might experience temporary dry mouth, thirstiness, or digestive symptoms (such as gas, cramping, or loose stools), as they might with salty or high fiber foods.

The principal investigator for the study was Dr. L. Cheskin, Dept. of Gastroenterology, Johns Hopkins School of Medicine.

Participants were instructed to be seated in the theater at least one seat apart from other participants, to eat and drink as much or as little of their chips and beverage as they desired, and not to share with anyone else. The theaters were monitored by several study staff during the movies. At the conclusion of the movie, participants clipped their bags of chips shut; noted the approximate amount of beverage consumed; and completed a brief questionnaire about product acceptance, satiety, and sensory attributes. Participants turned in the completed questionnaires and bags with uneaten chips and were given a toll-free telephone number to call if they had any questions or problems. Bags of chips were subsequently weighed to determine the amount consumed by each subject.

Trained telephone interviewers contacted study participants and administered a recall questionnaire to collect information on any effect experienced since the movie. All subjects were specifically asked if they had experienced any GI symptoms during or since the movie, and to specify those symptoms including the severity and timing of any such symptoms. The study protocol specified that participants be contacted within 2 to 4 days of viewing the movie. The petitioner reported that 85 percent were contacted within 2 to 4 days and a total of 97 percent were contacted within a week of viewing the movie.

The petitioner reported that the median consumption of olestra chips was approximately 2.1 oz (approximately 16 g of olestra)

compared to about 2.7 oz of triglyceride chips. Overall chip consumption was similar across age groups, but males generally consumed more chips than females (median of 2.8 versus 2.1 oz, p<0.01). The overall palatability of the triglyceride chips was rated slightly higher than the olestra chips, with a mean score of 6.4 versus 5.6 on a 9-point preference scale (p<0.01). Regarding satiety, there were no significant differences between the groups as indicated by mean satiety scores of 5.9 versus 5.7 for triglyceride chips and olestra chips, respectively, on a 9-point scale, with 9 being “extremely full” (p=0.07). Nor were any differences seen in beverage consumption, choice of beverage, or time since last meal prior to the movie between the two groups.

In 1996, FDA estimated the probable life-time averaged intake of olestra at the 90th percentile to be 7.0 g/p/d. To evaluate subchronic conditions, FDA estimated that a “high” acute consumer of olestra (everyday for 12 weeks) would consume 20 g/p/d, equivalent to eating a 2 oz bag of potato chips every day, and the 99th-percentile single-day intake of olestra for the group consuming the highest level of savory snacks to be 45 g/d (61 FR 3118 at 3124).

The petitioner attributed the lower chip consumption in the olestra group to the slightly lower preference for olestra chips reported by study participants. The petitioner stated, however, that the median consumption (2 oz) was more than a typical single-serving snack size bag of chips, and that approximately 100 participants ate more than 4 oz of olestra chips (approximately 32 g of olestra).

The petitioner reported that the proportion of subjects who reported GI symptoms after consuming olestra chips was not different from that after consuming triglyceride chips (15.8 percent and 17.6 percent respectively). There were no differences between the olestra and triglyceride groups in the frequencies for 14 different self-reported GI symptoms (gas, diarrhea, pain, cramping, upset stomach, loose stools, nausea, bloating, indigestion, aftertaste, eructation, constipation, vomiting, bloody stool), overall symptom severity for any GI event, nor time to onset or duration of symptoms. The petitioner also reported that consumption levels did not correlate with the rate of symptom reporting in either the olestra or triglyceride group.

The petitioner planned to have 1,400 participants in the study and anticipated symptom reporting to be 10 percent for the triglyceride group and 15 percent for the olestra group. Using these assumptions, the study would provide 80 percent power for detecting

a 5 percent difference in the proportions of symptoms between the olestra and triglyceride groups. The final number of participants to complete the study was 1,092,

which was fewer than planned. The rate of symptom reporting in the triglyceride group was 17.6 percent, which was higher than planned. Given the actual number of participants, and the actual rate of reports of symptoms in the triglyceride control group, FDA found that the study had an 80 percent chance of detecting a 7 percent difference between the test groups (p=0.05) (Ref. 13).

Of the 1,742 individuals originally enrolled for the study, 1,123 kept their appointments. Thirty-one individuals could not be contacted for followup, leaving a total of 1,092 evaluable subjects.

FDA observed that 962 participants completed a post-movie interview within the 2 to 4 days goal of the study. Of the remaining 130 participants who were contacted, 124 participants were contacted in 5 to 10 days, 3 on the day of the movie, 1 within a day, and 2 within 23 days. FDA noted that P&G included data from these 130 participants in its analysis to enhance the sensitivity of its analysis.

FDA noted that in both the olestra and triglyceride groups, the most frequently reported GI symptoms were abdominal pain, diarrhea, and flatulence. These symptoms are also the symptoms most commonly reported to P&G's passive surveillance program. FDA agrees with the petitioner that in this study, there was no difference in the rate or severity reported for loose stools or abdominal cramps between subjects who ate olestra-containing chips and subjects eating triglyceride chips. FDA examined the percent of subjects reporting at different levels of chip consumption and found that reports of diarrhea increased for both the olestra and triglyceride groups with increasing consumption of chips, but there was no difference in the rate of reporting between the groups (Refs. 10 and 13).

3. Home Consumption Study

The Home Consumption Study

was designed to measure, under market use conditions, the effect of eating chips made with olestra on GI symptoms in adults and children over an extended period of time (Ref. 14). This double-blind placebo-controlled trial represented 1,138 households (3,181 individuals, ages 2 to 89) randomly assigned to either the olestra group or the control group. To be enrolled, at least half the members of the household had to have eaten corn or potato chips at least four times in the previous month and all members of the household had to be willing to participate in the 6-week long study. A contact for each household was identified and was required to return to the study site once a week for 6 consecutive weeks. During each visit, the contact could choose from a selection of potato chips and tortilla chip products labeled as containing either olestra or triglyceride. The selection of snacks used in the study were products available in the marketplace presented in typical packaging. To encourage snack consumption, up to eight bags of chips (varying in weight from 5.5 to 9 oz) could be selected each week. For the households in the olestra group, the olestra-labeled packages contained olestra chips, but for the control group, the olestra-labeled packages contained triglyceride chips. For both groups, the triglyceride-labeled packages contained triglyceride chips. All olestra-labeled products displayed the olestra label statement.

Dr. R. Sandler, Professor of Medicine, University of North Carolina, Chapel Hill, was the principal investigator for this study.

At each weekly visit, the contact would also provide daily records kept by each member of the household regarding GI symptoms. The household contact assisted and/or completed the form for children. The record consisted of a check list of eight specific GI symptoms

as well as a field to write in any other symptoms. On each day a GI symptom was recorded, the subject was to rate the effect of those symptoms on daily activity using a scale ranging from “noticed but did not affect” to “missed all day at work/school.” Medication use and physician visits were also to be recorded.

The list of GI symptoms include the following descriptions: (1) Heartburn or indigestion, (2) nausea or queasiness, (3) vomiting, (4) gas, (5) bloating, (6) abdominal cramping or pains, (7) more frequent bowel movements, and (8) looser stool.

There were 1,620 subjects from 568 households in the olestra group and 1,561 subjects from 570 households in the control group. The groups were similar with respect to age, sex, and race. Subjects ate chips frequently throughout the study. The median number of days on which a subject consumed an olestra-labeled chip was 20 days of a possible 42 days for the olestra group and 21 of a possible 42 days for the control group. The length of the study and the large number of individuals per group resulted in a collective period of more than 30,000 “eating” days, making it possible to detect small differences in the reporting of GI symptoms. The median total amount of olestra-labeled chips eaten over the course of the study by the olestra group (25.2 oz) was slightly less than that eaten in the control group (27.6 oz). During the 42-day study, subjects whose consumption was in the top 10 percent of the olestra group ate more than 59 oz of chips, while in the control group, the top 10 percent of the group ate more than 70 oz of chips. The petitioner presented data to show that the rates of olestra consumption achieved were beyond customary snacking by comparing the intake of olestra at the 90th percentile of consumption in this study (13.3 g/d) to Market Research Corp. of America (MRCA) preapproval estimates (6.4 g/d), and to data collected regarding “real-world” olestra consumption in the Active Surveillance Study (2.1 g/d). The petitioner concluded that the rates of consumption achieved in this study for both the olestra and triglyceride groups were higher than usual snack consumption.

The petitioner reported that for its original planned analysis for the study, which examined the percentage of eating days where GI symptoms were reported within 2 days, olestra-containing chips resulted in an increase (p<0.05) in the GI symptoms of more frequent bowel movements, loose stools, and gas. There was no increase in reports of abdominal cramping or any of the other individual or total GI symptoms. The petitioner decided that this analysis could not be clearly interpreted because olestra labeled chips were eaten on numerous days of the study and therefore a particular GI event would be associated with 2 or 3 eating days.

The petitioner presented data from an analysis that compared the occurrence and frequency of GI symptoms between the olestra and control groups. The primary response variable was the percentage of individuals reporting a GI event. For all subjects who consumed olestra-labeled products, the petitioner found that there was no difference in the total percentage of subjects reporting a GI symptom between the olestra and control groups. Of the eight GI symptoms evaluated, the only difference was an increased number of reports of nausea for the control group. For those subjects who reported a GI event, the number of symptom days was also compared. When the petitioner examined the data by days on which subjects reported symptoms, there was a small increase in the olestra group in the number of days when more frequent bowel movements were reported (3.7 days for olestra compared to 2.8 days for controls; p=0.04). The petitioner calculated that this increase was about one symptom day out of the 42 days of the study. The petitioner reported that

subjects' self assessments showed little or no impact of GI symptoms on subjects' daily life, and there was no increase in the percentage of reported severe impacts in the olestra group compared to the control group.

The petitioner also examined whether there were differences in the incidence of reported GI symptoms among the different age groups. The petitioner reported that there were no significant differences in total or specific GI symptoms between the olestra and triglyceride groups for children (2 to 12 years; n=885), teens (13 to 17 years; n=227), or the elderly (65 to 89 years; n=402), even among the highest consumers. This analysis showed that for adults (18 to 64 years; n=1667), there was an increased percentage in reporting the GI symptom gas in the olestra group compared to the control group. There was also an increase in the number of GI symptom days, and an increase in the number of more frequent bowel movement symptom days, among adult subjects eating olestra of approximately one symptom day out of the 42 days of the study.

Among adult females in the olestra group, compared to adult females in the control group, there was an increase of approximately one symptom day out of 42 days of the study with regard to more frequent bowel movements, gas, and any GI symptom. The only difference regarding reports of abdominal cramping was an increase in the control compared to the olestra group for adult males.

The petitioner concluded, based on the subjects' self assessments, that none of these reported increases in the number of symptom days were meaningful because there was no impact on subjects' daily activities. Based on its comparison of the percent of subjects who reported one or more GI events during the course of the study, P&G concluded that there were no meaningful or serious GI effects associated with eating olestra-containing chips.

At the end of the 6-week study, P&G asked participants which kind of chips they thought they were eating from the olestra-labeled bags. P&G reported that the percentage of subjects reporting GI symptoms was greater (approximately 50 percent) in those who believed they were eating chips made with olestra compared to those who thought they were eating triglyceride chips. This was true regardless of whether the participant was actually eating olestra or triglyceride chips.

FDA employed a number of statistical approaches to best address the different questions to be answered by the study, and while such differing approaches may yield different answers, this varied approach provides a more complete picture of the study results. FDA analyzed both the temporal relationship between consumption and symptoms, and summation data for the study (Refs. 15 and 16).

Examination of temporal data is important for evaluating an association between olestra intake and GI symptoms. Such an analysis is also important because in a study of this length, subjects can modify their eating behavior based on their experience with a product. FDA found that subjects in both the olestra and triglyceride groups modified their intake of chips as a result of experiencing more frequent bowel movements. FDA was able to conclude that consumers modify their behavior based on their experience with olestra chips by examining the amount of chips consumed the day before, the day of, and the day after a report of more frequent bowel movements. Chip consumption decreased after experiencing more frequent bowel movements, although consumption of chips did not cease.

In order to understand the temporal relationship between olestra consumption and GI symptoms, FDA examined the frequency of GI symptoms for numerous different patterns of olestra consumption over a period of several days.

In all these analyses, FDA found that for men, olestra consumption resulted in an increase in any GI symptom, gas, and more frequent bowel movements, and a decrease in nausea. For women, olestra consumption resulted in an increase in any GI symptom, gas, looser stools, and more frequent bowel movements. On the day that chips were eaten, the difference in the percentage of occasions that more frequent bowel movements were reported for the olestra chips compared to the triglyceride chips was 1.6 percent for males and 1.2 percent for females. These effects were seen on days of consumption of olestra chips but not on subsequent days on which olestra-containing chips were not eaten. When olestra chips were consumed on consecutive days there was some cumulative effect for the reports of these GI symptoms. This was particularly true for males. For example, the difference in the percentage of occasions that a report was made in the category “any GI symptom” for the olestra chips compared to the triglyceride chips increased from 0.9 percent on the first day to 1.7 percent on the second day, to 2.6 percent on the third consecutive day that chips were eaten and a complaint was recorded. There was also a trend for more frequent and recent consumption of olestra to result in a GI symptom. While increasing consumption of olestra and triglyceride chips both resulted in more symptoms, the effect of olestra was greater compared to triglyceride chips at all doses.

These included determining the percent of occasions for which GI symptoms occurred on the same day and the following 2 days of eating an olestra-labeled chip; comparing the frequency of occurrence of GI symptoms on days that olestra-labeled chips were eaten to days that chips were not eaten to determine a “same day of eating effect”; determining the percent of days on which GI symptoms were reported for all non-eating days in order to evaluate possible delayed or continuing effects of olestra; comparing the percent of days on which GI symptoms were reported to the number of consecutive days eating olestra-labeled chips in order to examine possible cumulative effects; for various GI symptoms analyzing the pattern of consumption of olestra-labeled chips for the days prior to the GI symptom in order to examine how the most recent day of eating and the frequency of eating is related to the GI symptom; for various GI symptoms, determining the amounts of olestra-labeled chips consumed on the day the GI symptom occurred (Ref. 16).

In examining the effect of olestra consumption on different age groups, FDA found that GI symptoms were primarily seen in the 18 to 64 age group. There were no olestra-related effects in the groups over 65 years or younger than 18 years.

In a separate statistical analysis, FDA focused on the sum total of symptom days and consumption of olestra-labeled chips over the course of the 42-day study (summation data). FDA analyzed the data for each GI symptom for both the entire study population, and for a population divided based on age and gender.

In the statistical analysis of the sum total of symptom days over the course of the 42-day study, FDA first examined the relationship between the reporting of particular GI symptoms and the consumption of olestra-containing foods by comparing the olestra group and the triglyceride group. FDA found that for all study subjects (males and females) over the course of the 42 day study, there was an increase of 0.28 more frequent bowel movement symptom days in the olestra group compared to the triglyceride group. FDA then examined the relationship between the reporting of particular GI symptoms and the consumption of olestra-containing foods by analyzing the olestra group and the triglyceride group separately by gender. FDA found for females in the olestra group, there was an increase in “any GI symptom” of 0.5 mean symptom days compared to the females in the triglyceride control. It was also observed that for females in the olestra group, there was an increase of 0.3 symptom days in more frequent bowel

movements over the course of the 42 day study compared to females in the triglyceride group. For males in the olestra group, the analysis showed an increase of 0.24 more symptom days for more frequent bowel movements compared to males in the triglyceride group.

FDA then examined the relationship between the amount of product consumed and symptoms reported for all study subjects (males and females), and found there were associations between olestra consumption and reports of “any GI symptom” (p=0.03), loose stools (p=0.006), and more frequent bowel movements (p=0.002). No such associations were observed between the consumption of the control chips (triglyceride chips labeled as olestra) and any measured symptom. When analyzed separately by gender, both sexes showed trends for an association between the consumption of olestra and loose stools (males p=0.001, females p=0.018), and more frequent bowel movements (males p=0.001, females p=0.042), but only males also showed a trend for an association between the consumption of olestra and “any GI symptom” (p=0.001).

FDA examined the relationship between the consumption of olestra-containing foods and reports of abdominal cramping. FDA found no difference in the frequency of reported abdominal cramping between the olestra group and the triglyceride group. FDA analyzed the olestra and triglyceride groups separately by gender for reports of abdominal cramping and found no difference between males or females in the olestra group as compared to the triglyceride group. FDA agrees with the petitioner that there was no observed difference in the incidence or association of reported abdominal cramps between the olestra group and the triglyceride group (Ref. 10).

4. Stool Composition Study

The Stool Composition Study was sponsored

by the petitioner as a followup to the preapproval Fecal Parameters Study (discussed previously in section II.A.2 of this document). The study was designed to establish whether consumption of olestra-containing foods is associated with changes in clinical measures of diarrhea (water and electrolyte loss), effects which may be harmful, or stool consistency alone, which may result from adding bulk to the stool and which is not harmful. In addition, the study was designed to determine the relationship between objective measures of clinical diarrhea (

e.g.

, stool water output and bowel movement (BM) frequency) and subjective reports of “diarrhea” from study subjects. The effects of olestra were compared to a placebo, triglyceride chips, and to sorbitol, an osmotically active sugar alcohol that was chosen as a positive control to ensure that the study methodology was adequately sensitive to detect increases in stool water output.

The GI consultant to the study was Dr. R. Gianella, University of Cincinnati.

The study was a single-site, randomized, double-blind, placebo-controlled parallel clinical trial. Sixty-six subjects, ages 18 to 74, were housed on a metabolic ward for 12 days and consumed meals ad libitum. The meals conformed to the American Heart Association Step I diet guidelines (no more than 30 percent of calories from fat). Beverages were available ad libitum. All study subjects had to consume 5 oz of potato chips eaten as two afternoon snacks. A serving of potato chips was either olestra (test) or triglyceride (placebo). All subjects were also required to consume 1.5 oz of candy made either with sorbitol (test) or sucrose (placebo) as a morning snack. The first two days (study days 1 and 2) were a lead-in period during which subjects were acclimated to the living conditions and the diet, and consumed placebo snacks (triglyceride potato chips and sucrose candies). Stool samples were not collected during the lead-in period. The next 4 days (study days 3 to 6) comprised the baseline period, in which subjects continued to consume placebo snacks, and all stool samples, BM ratings, and GI symptoms were collected. For the final 6 days (study days 7 to 12), subjects consumed snacks according to their randomly assigned treatment group, and all stool samples, bowel movement ratings, and GI symptom reports were collected. There were two olestra test groups (20 g and 40 g olestra) and two control groups (positive control of 40 g sorbitol and placebo). The placebo group consumed two servings of placebo (triglyceride) potato chips and placebo (sucrose) candy. The positive control group consumed two servings of placebo potato chips and test candy (40 g sorbitol). The 20 g olestra test group consumed one serving of test potato chips (olestra), one serving of placebo potato chips, and placebo candy. The 40 g olestra group consumed two servings of test chips and placebo candy.

The petitioner noted that in the study the doses of olestra were threefold to sixfold more than the estimated daily intake, and 10 to 20 times more than the observed intake at the 90th percentile level in the Active Surveillance Study (see section III.D.1 of this document). The high dose, 40 g/d, was higher than the highest dose used in the preapproval nutrition studies (32 g/d) described previously in section II.A.3 of this document, in which the high dose group experienced an increase in GI symptoms, specifically in reported diarrhea/loose stools. In that preapproval study, FDA concluded that the reported diarrhea was not diarrhea in the medical sense because there was no evidence of subjects experiencing significant fluid or electrolyte loss (hemoconcentration, electrolyte imbalance; 61 FR 3118 at 3152-3154).

The petitioner concluded that with regard to the critical parameters that are medically relevant in defining diarrhea, the objective measures showed that olestra did not meaningfully change either the total stool output or stool water output, while sorbitol produced large effects on both parameters. Compared to baseline, mean stool water output increased 9 g/d and 37 g/d for the 20 and the 40 g/d olestra groups respectively, and 325 g/d for the 40 g/d sorbitol group. Stool water output decreased 28 g/d for placebo. The measured mean stool water content for the sorbitol group was nearly 10 times greater than the group consuming the highest level of olestra and the number of watery BMs was 140 in the sorbitol group, one in the 40 g/d olestra group, none in the 20 g/d, and one in placebo. While sorbitol significantly increased the severity of abdominal cramping compared to placebo, olestra did not. The petitioner found that olestra consumption did not result in any clinically meaningful increases in objective measures of diarrhea, namely, total stool output, bowel movement frequency, and stool water and electrolyte output. The mean number of BMs for the olestra 40 g/d group was increased compared to placebo but was not increased compared to the olestra 20 g/d group. Subject reports of “watery, difficult to control diarrhea” did not necessarily correlate with measured viscosity of the stool. Olestra did increase stool weight in proportion to the amount eaten, and daily consumption of olestra gradually softened stool in a dose-responsive manner. The sponsor found that there was increased reporting of “diarrhea” in the olestra treatment groups during the treatment phase without an increase in total water output outside the normal range,

i.e.

, the range observed during the baseline period and in the placebo group.

P&G concluded, based upon the study results, that the consumption of olestra

does not cause diarrhea, but simply adds bulk and softening to the stool.

FDA reviewed the data from this study and agrees with the petitioner's analysis, although some of the agency's analytical strategies differed from those of the sponsor (Ref. 17). FDA concludes that both comparisons of the mean after treatment and of changes from baseline showed dose responsive increases in stool characteristics (total output, water output, consistency, frequency and increases in water content) that were not clinically significant (Ref. 18).

Using a 7-point scale to rate consistency of bowel movements (1 = watery, diarrhea; 4 = normal; 7 = hard, constipation), subjective ratings of stool consistency showed that subjects who ate 40 g/d olestra perceived their stools to be looser (mean rating 2.4) compared to those who ate 20 g/d olestra (mean rating 3.1). By comparison, placebo subjects had a mean score of 3.9 whereas those subjects in the 40 g/d sorbitol group had a mean score of 1.5 (mean scores determined for days subjects consumed snacks according to their randomly assigned treatment group). When stool consistency was measured by peak force value for extrusion, both olestra groups had a lower mean stool consistency than placebo and the 40 g/d olestra group was lower than the 20 g/d group. These dose responsive findings seen among subjects eating olestra resulted from gradual stool softening effects observed after several consecutive days of olestra consumption. Although subjects characterized these viscosity changes as “diarrhea,” the changes were not associated with an increase in stool water.

FDA examined the percentage of symptom days for cramping and found that although the 40 g/d olestra group reported an increased incidence of abdominal cramping compared to those in the 20 g/d olestra group (35.8 percent compared to 9.8 percent), this difference did not rise to statistical significance. The percentage of subjects reporting abdominal cramping in the 20 g/d olestra group appeared to decrease when compared to baseline or placebo (9.8 percent compared to 20.5 percent or 18.3 percent). The 40 g/d sorbitol group had the highest percentage (69.8 percent) of reports of cramping. Subjects rated symptom severity on a scale of 0 to 5, with 0 representing none and 5 extreme. The severity of cramps reported by subjects in the olestra 40 g/d group was less severe than that reported by subjects in the 40 g/d sorbitol group (0.72 compared to 2.3). No significant olestra effects were found for GI symptom severity, although one individual in the 20 g/d olestra group reported severe urgency at a rating higher than any other report in any of the other groups (Refs. 10 and 18).

5. Comments Regarding the GI Studies

FDA received comments about the new GI studies. FDA considered these comments and responds in the following paragraphs. Comments regarding the label statement for GI effects will be discussed in section V of this document.

(Comment 5) A comment from CSPI to the current petition criticized the Rechallenge Study. The comment stated that the study subjects were not screened for sensitivity to olestra, as was done in the preapproval Fecal Parameters Study. The comment also asserted that the Rechallenge Study contained a strong likelihood of bias because only 10 percent of those contacted agreed to participate in the rechallenge and those that did participate consumed olestra on only 2 days, at least 1 week apart, which reduced the sensitivity of the study. CSPI asserted that the Rechallenge Study also assumed that those sensitive to olestra would respond to it 100 percent of the time. The comment contended that those experiencing adverse reactions may only do so under certain circumstances, not 100 percent of the time.

FDA does not agree that the selection of study subjects biased the Rechallenge Study nor does CSPI provide such evidence. FDA has determined that the subjects who participated in the Rechallenge Study were adequately representative of those persons who contacted the postmarketing surveillance system in terms of severity of initial symptoms and amount of olestra reportedly consumed prior to the initial symptom episode (Ref. 11). Further, CSPI provided no basis for its assertion that additional subject screening is necessary to accomplish the objectives of the Rechallenge Study.

CSPI states that the sensitivity of the Rechallenge Study was reduced because participants consumed olestra on only 2 days, at least 1 week apart. The conditions of the study were designed to be similar to the conditions under which the subjects originally reported effects that they attributed to consuming an olestra-containing snack. FDA found that for nearly three-quarters of the subjects, the amount of olestra consumed in the study was comparable to, or greater than, the amount associated with their initial symptom episode (Ref. 11). In addition, more than three-quarters of the subjects reported that their initial symptom episode occurred after a single eating occasion. Therefore, subjects were challenged with 2 oz of olestra chips on two occasions separated by a week, providing a dose and number of exposures comparable to, or greater than, those associated with many of the subjects' initial symptom episodes.

CSPI's comment did not reference where FDA or the petitioner assumed that those sensitive to olestra would respond to it 100 percent of the time, nor is FDA aware of anyone who has put forth such a position. Indeed, FDA agrees that even if an individual experiences a reaction to olestra, that individual may not experience such reaction after every exposure. The Rechallenge Study shows that subjects exposed to olestra containing-chips were no more likely to report GI symptoms than when exposed to an equal amount of triglyceride chips. Thus, the study subjects' reactions to olestra containing-chips are not so frequent that they can be distinguished from their reactions to regular chips under the conditions of the test.

(Comment 6) A comment from CSPI to the current petition criticized the Acute Consumption Study. CSPI's comment relies on its published letter

commenting on a published study (Ref 12.) that reports data from the Acute Consumption Study. CSPI stated that the study may have failed to detect the true incidence of GI effects due to a lack of statistical power or inadequate controls. For example, with the incidence of “any GI event” of about 15 percent, 550 subjects in each group would have provided only about a 50 percent probability of detecting a 5 percent actual increase in the treatment group. Along the same lines, diarrhea and loose stools were increased less than 1 percent in the olestra group compared to baseline levels of 2.6 percent and 1.1 percent, respectively. The comment asserted that maintaining 80 percent power to detect a 1 percent increase over a 2 percent baseline requires about 4,000 subjects per group. The comment also contends that the darkened movie theater may potentially cause exposure misclassification (some “olestra eaters” may have eaten few or none of their chips; some “non-olestra eaters” may have eaten friends' olestra chips). The comment also stated that it took up to 10 days after consumption to assess symptoms. CSPI also pointed out that non-olestra eaters consumed one-third more chips than the olestra eaters.

CSPI's published letter was included in the comment as an attachment (Ref. 19).

The criticism by CSPI of the Acute Consumption Study does not negate the

conclusion that FDA reached in its analysis of the study. The Acute Consumption Study was conducted to provide information relevant to whether olestra-containing foods should bear a label statement that informs consumers about the potential GI effects associated with olestra. FDA points out that the Acute Consumption Study was only one of several studies under consideration in this petition, and that the agency's decision on the petition is based on the totality of evidence in the record.

While the petitioner's Acute Consumption Study did not achieve the statistical power that P&G originally desired (80 percent power to detect a 5 percent difference between treatment groups), the study still provides meaningful information concerning the effect of olestra-containing foods on the GI system. FDA's scientific review determined that the study does have 80 percent power to detect a 7 percent difference between treatment groups (Ref. 13). The study showed that there was no difference in the rate or severity of loose stools or abdominal cramps between subjects who ate olestra-containing chips compared to those who ate triglyceride-containing chips.

The comment provides no evidence that the darkened theater or the method used to collect symptom data affected the outcome of the study. As discussed previously, the study protocol was designed to minimize the possibility of inaccurate measurements or subjects' sharing of chips. For example, study participants were instructed to be seated in the theater at least one seat away from other participants and not to share their chips or beverage with anyone else. The theaters were also monitored by several staff during the movie.

Similarly, the comment did not explain the effect on the study results, if any, from the 10-day period used to assess symptoms. After the movie, trained telephone interviewers contacted study participants and administered a recall questionnaire to collect information on any effects experienced since the movie. The study protocol specified that participants be contacted within 2 to 4 days of viewing the movie. The petitioner reported that 85 percent of study subjects (962 of the 1,092) were contacted within 2 to 4 days of viewing the movie, an additional 124 subjects were contacted in 5 to 10 days.

Of the remaining subjects, three were contacted on the day of the movie, one within a day, and two within 23 days (Ref. 13).

FDA agrees that the median chip consumption for the control group was greater than that for the olestra group. As discussed previously, the Acute Consumption Study was designed to be an ad libitum study, allowing the investigators to examine the effects of customary or usual consumption. As an ad libitum study, it is possible that one group of subjects may consume more chips than the other. For example, the median consumption of chips made with olestra was 2.1 oz compared to 2.7 oz for chips made with conventional triglycerides. CSPI did not explain how the fact that one group of subjects ate more chips than the other affects the conclusions drawn from this study regarding the need for special labeling.

(Comment 7) A comment from CSPI to the current petition criticized the Home Consumption Study. CSPI's comment relies on its published letter

commenting on a published study that reports data from the Home Consumption Study (Ref. 14). The comment raises five issues: (1) The comment stated that some of the data relating to the highest decile of olestra consumers were overlooked; (2) the comment argued that it is important to focus on the small number of heavier consumers because most subjects ate relatively few olestra-containing chips; (3) the comment stated that in the highest decile of olestra consumers the incidence of more frequent bowel movements and loose stools was twice that of controls; (4) the comment stated that olestra consumers in the highest decile had symptoms on 18 percent of person-days, compared to 12 percent of person-days in the control group (table 4 in Ref. 14); and (5) the comment pointed out that olestra consumers missed some or all of their activities on 0.4 percent of days, compared to 0.2 percent in the control group.

CSPI's published letter was included in its comment as an attachment (Ref. 20).

Prior to publication of the article concerning the Home Consumption Study (Ref. 14), FDA conducted its own indepth analysis of the raw data from the Home Consumption Study (Refs. 15 and 16) and described this analysis at the 1998 FAC meeting in which CSPI participated. FDA's analysis included an estimate of the extra symptom-days experienced by subjects in both the 90th and 95th percentile of olestra-containing chip consumption (Ref. 15). Subjects at the 90th percentile ate 64 oz of olestra-containing chips over the course of the study while those at the 95th percentile ate 83 oz of olestra-containing chips over the course of the study. Although CSPI alleges that the subjects in the study ate relatively few olestra-containing foods, the petitioner presented data to show that, in fact, the rates of olestra consumption achieved in the study were beyond usual snack consumption.

As part of the Home Consumption Study, the investigators considered the effect of GI symptoms on subjects' daily activities. In its comment, CSPI points out that olestra consumers missed some or all of their activities on 0.4 percent of days, compared to 0.2 percent in the control group, implying that this is significant. FDA disagrees.

CSPI does not explain how it calculated the percentage of days on which subjects missed some or all of their activities, nor does CSPI provide statistical analyses to assess whether these differences occurred by random chance (

e.g.

, illness unrelated to olestra). FDA was able to replicate the numbers that CSPI presented and performed tests of statistical significance on the data. The actual number of days on which subjects in the highest decile missed some or all of their activities is very small (9 of 2,226 days in the olestra group versus 5 of 2,646 days in the control group). Five subjects in the olestra group and four subjects in the control group missed some or all activities at least 1 day. The number of subjects missing activities and the number of days missed by these subjects are comparable for the olestra and control groups, except for one subject in the olestra group who missed some or all activities on 4 days (Ref. 21).

From these data, it cannot be concluded that for the highest decile of consumers olestra consumption resulted in an increase in days in which consumers missed some or all activities. FDA believes that the Home Consumption Study, designed to examine the effects of “real life” olestra consumption, provides useful information relevant to the labeling of olestra-containing foods. CSPI does not show how their analysis would change FDA's conclusions.

Sophisticated statistical models are impractical for such a small number of cases. However, a Fisher's Exact test showed that the proportion of subjects in the olestra group who missed some or all activity at least 1 day was not significantly different (p-value of 0.73) from the proportion of subjects in the control group who missed some or all activity at least 1 day.

(Comment 8) A comment from CSPI to the current petition criticized the petitioner's Stool Composition Study. The comment stated that this study does not negate and should not supersede the two preapproval 8-week studies or the preapproval Fecal Parameters Study. In its comment, CSPI cites a 1995 FDA memorandum discussing the Fecal Parameters Study (Ref. 22) and asserts that the memorandum says that several

subjects in the study experienced high rates of water loss through their stool. The comment also stated that the definition of diarrhea used in the Stool Composition Study was too narrow and is not consistent with the definition used by the Centers for Disease Control and Prevention (CDC; three or more loose stools in a 24 hour period). The comment asserted that self-reporting is usually considered sufficient to conclude that people experience diarrhea regardless of demonstrated loss of electrolytes.

In contrast to the Acute Consumption Study, the Home Consumption Study, and the Rechallenge Study, the Stool Composition Study was designed to extend the understanding of olestra's effect on stool characteristics that would potentially represent a safety concern. For this reason, the Stool Composition Study was conducted under conditions most likely to elicit GI effects. The highest dose of olestra provided in the Stool Composition Study (40 g/d) was greater than the 32 g/d used in the preapproval 8-week studies which was shown to cause an increase in GI symptoms (specifically in reported diarrhea/loose stools) and was twice as high as the highest dose given in the preapproval Fecal Parameters Study (20 g/d). Additionally, subjects' stool samples were collected for all 6 days of the treatment period in the Stool Composition Study, compared to only three days of the 7-day treatment periods in the Fecal Parameters Study. The Stool Composition Study does not negate the preapproval studies, but the results of the preapproval studies must be considered in light of those from the Stool Composition Study.

The results of the Stool Composition Study show that olestra consumption does not result in any clinically meaningful increases in the objective measures of diarrhea. Importantly, the Stool Composition Study assessed the effects of olestra consumption using objective parameters such as total stool output, bowel movement frequency,

and stool water and electrolyte output rather than a subject's subjective assessment of whether he or she experienced diarrhea. The use of objective measures of diarrhea is necessary to assess whether the “diarrhea” experienced by study subjects represents a safety concern.

FDA notes that the mean bowel movement frequencies in the olestra-consuming groups were less than three bowel movements per day. The mean bowel movement frequencies were 1.6 ± 0.2 BM/d (mean ± standard error) in the 20 g/d olestra group and 2.0 ± 0.2 BM/d (mean ± standard error) in the 40 g/d olestra group (Ref. 18).

FDA was concerned with the potential for olestra to cause diarrhea because diarrhea of medical significance is associated with excessive water loss and electrolyte loss, which may raise safety concerns. The Fecal Parameters Study memorandum cited by the comment states that the stool water concentration of subjects who reported having diarrhea during the olestra 20 g/d period did not differ from that of their nondiarrheal stools during the placebo period. The memorandum also states that although the percent of water in the stools may not have differed, it is possible that absolute water loss was greater in subjects reporting olestra-associated diarrhea because of the greater mass (weight) of stool passed. FDA concluded in the 1996 final rule that the loose stools experienced in the preapproval clinical studies were not diarrhea in the medical sense because they were not associated with loss of water or electrolytes (61 FR 3118 at 3159). The agency also stated that even those subjects in the 8-week studies who experienced loose stools or diarrhea continuously for several weeks during olestra consumption did not show any evidence of fluid loss such as hemoconcentration or electrolyte imbalance. Thus, the agency determined that olestra-related GI effects were not adverse health effects (61 FR 3159). The results of the Stool Composition Study confirm the agency's 1996 decision that the GI effects resulting from olestra consumption do not represent adverse health effects, regardless of the terminology (diarrhea or otherwise) used to describe these effects.

D. A Study Regarding Nutritional Effects—Active Surveillance

As discussed previously in section II of this document, olestra is neither digested nor absorbed, and as such, passes intact through the digestive tract where it can interact with fat-soluble dietary components present in the gut at the same time. Fat-soluble nutrients and components tend to partition or dissolve into the olestra, thereby reducing the absorption efficiency of these substances (61 FR 3118 at 3144-3149). Olestra does not interfere with the absorption of macro-nutrients (protein, carbohydrates, and fats) or water-soluble nutrients (61 FR 3118 at 3149-3152). The clinical studies conducted in support of the 1996 final rule examining the effect of olestra on fat-soluble components of the diet were performed under conditions that maximized the interaction of olestra with these dietary components,

i.e.

, olestra was incorporated into foods eaten at every meal. These studies were not designed to examine effects from the usual or customary consumption of savory snacks made with olestra (

see

section II.A.3 of this document).

To compensate for the effect of olestra on the absorption of the fat-soluble vitamins A, D, E, and K, FDA required that these vitamins be added to olestra-containing foods. The level of addition was chosen to ensure that there would be neither a reduction in the absorption of fat-soluble vitamins from the diet, nor an increase in vitamin levels due to the presence of the added vitamins in the olestra-containing foods (see section II.A.3 of this document). Although FDA noted that olestra interferes with the absorption of carotenoids, FDA found no scientific basis for requiring the addition of any carotenoid to olestra-containing foods (61 FR 3118 at 3147-3149).

As outlined by the petitioner in its January 24, 1996, letter to the agency, P&G established a program of active surveillance. A report of this surveillance with results and analysis from the first year at the sentinel site was submitted to the agency on April 15, 1998. Additionally, the agency has continued to review and evaluate new data and information that bear on the safe use of olestra, such as new data and information on the health significance of carotenoids.

1. Active Surveillance Study by P&G

The petitioner provided funding to investigators at the Fred Hutchinson Cancer Research Center in Seattle, WA, to design and implement a multi-year, Active Surveillance Study to monitor patterns of use of olestra-containing savory snack products and to collect blood samples to measure nutrient status (Ref. 23). The study had three specific goals: (1) To monitor adoption and patterns of use of olestra-containing savory snack products in representative samples of the U.S. population; (2) to assess the association between the introduction of olestra-containing savory snacks and serum concentrations of carotenoids and fat-soluble vitamins in representative cross-sectional samples of the U.S. population; and (3) to assess the long-term association between consumption of olestra-containing savory snacks and serum concentrations of carotenoids and fat-soluble vitamins among a cohort of olestra consumers.

The study has three components corresponding to the three specific aims. The first component, called the population cross-section, was a telephone survey used to monitor the prevalence and patterns of olestra-containing savory snack consumption,

fruit and vegetable consumption, and triglyceride savory snack food consumption by consumers. Demographic information was collected as well. A telephone survey was conducted in each of the study sites before olestra-containing snacks were marketed. Subsequent yearly surveys were completed after olestra-containing snacks were introduced to the market.

A random sample of participants in each telephone cross-section sample was recruited into the second component, a clinical cross-section. The clinical cross-section was an investigation of the relationships among nutrient intake, olestra consumption, and serum nutrients. Study participants visited a clinic to provide further information, including dietary information, medical histories, and blood samples. Followup telephone interviews included questions about usual fruit, vegetable, and snack food use during the previous month, a 24-hour dietary recall to measure co-consumption of fruits and vegetables with savory snacks, health symptoms and status, and a short household food inventory.

Within the clinical cross-section, information on olestra intake was used to select olestra users from non-users to be recruited into the third component of the Active Surveillance Study,

i.e.

, the clinical cohort study. The clinical cohort study was an investigation of the relationships among nutrient intake, olestra consumption, and changes in serum nutrients over time. Participants in the clinical cohort are a subset of those people who participated in the Year 0 clinical cross-section and were monitored annually over the course of the Active Surveillance Study. The clinical cohort was designed to have an over-representation of consumers of olestra-containing snack food. The design for the clinic visit and the information gathered is the same as for the clinical cross-section.

The study was conducted in four U.S. cities. As of the publication of this document, data are available only from the sentinel site, Marion County, IN, where test marketing began in 1997.

The study began one year later in the other cities, because national marketing of olestra-containing foods in those areas began later.

The other three cities in the Active Surveillance Study were Baltimore, MD, San Diego, CA, and Minneapolis, MN.

The first component of the active surveillance is the population cross-section. A random-digit-dial telephone survey of Marion County, IN, residents was completed before olestra-containing foods were marketed in that area (February 1997). This survey (Year 0) included 1,962 adults, aged 18 years and over. The second telephone survey was completed after olestra-containing foods were introduced to the local market (between August 1997 and January 1998). This survey (Year 1) included 1,525 adults, aged 18 years and over. Based on the Year 1 data, which are weighted to be representative of the Marion County population, 15.5 percent of adults reported eating olestra-containing snacks one or more times per month with the median frequency being three times per month. Ninety percent of adults reported eating one or more servings of fruits and vegetables per day, thus providing a basis for assessment of any effects on dietary carotenoid absorption. Intake of fruits and vegetables and intake of total snacks did not change in the population cross-section between Year 0 and Year 1. Olestra-containing snack food introduction was not associated with an overall increase in savory snack consumption or with a decrease in fruit and vegetable intakes. There was a modest decrease in consumption of reduced- and non-fat savory snacks at Year 1 compared to Year 0.

Blood sera from study subjects, in both the cross-sectional and clinical cohorts, were analyzed for vitamins A, D, E, and K, total cholesterol, high density lipoprotein (HDL) and triglycerides, and the six major carotenoids that represent more than 90 percent of the circulating carotenoids (alpha and beta carotene, lycopene, lutein, zeaxanthin, and beta-cryptoxanthin). The study investigators then compared these serum measures based on olestra intake. Four olestra consumption groups were defined: (1) None; (2) low (less than 0.4 g/d of olestra, which is less than the 60th percentile of consumption); (3) medium (between 0.4 and 2.0 g/d of olestra, which is between the 60th and 90th percentiles of consumption); and (4) high (greater than 2 g/d, which is greater than the 90th percentile of consumption).

Results from the cross-sectional study comparing 1,252 subjects in year 0, and 1,164 subjects in year 1, show that with increasing olestra intake, there were significant trends for an increase in vitamin K levels (p = 0.013) and a decrease in serum cholesterol (p = <0.05). There were no significant differences or trends found for other vitamins or for total carotenoid or individual carotenoids that could be associated with olestra consumption.

For the clinical cohort (477 study participants), the sponsor reported that for the entire cohort from year 0 to year 1, there was a decrease in mean serum concentrations of total carotenoids, as well as in concentrations of retinol, 25-OH vitamin D, lycopene, lutein, and zeaxanthin, and an increase in beta-cryptoxanthin. Tests of association between olestra consumption and changes in serum concentrations of fat-soluble vitamins and carotenoids were based on regression models that included variables to characterize the four levels of olestra consumption. However, these changes were not related to the amount of olestra consumed. A trend was observed for increased vitamin K, but the change did not reach statistical significance (p = 0.087). There were no changes observed for the other vitamins.

The petitioner cautioned that the results discussed previously reflect data from only the first year that olestra products were marketed, and that data were available from only a single site. With these caveats, the petitioner reached the tentative conclusion that it appeared that the consumption of olestra-containing foods in the marketplace had little, if any, effect on the status of fat-soluble vitamins and nutrients as measured by serum concentration.

FDA notes that survey results show that the co-consumption of savory snacks (made with or without olestra) with a fruit or vegetable was relatively rare. Overall, less than 15 percent of total carotenoids were consumed with any savory snack. Olestra's effect on the absorption of fat-soluble carotenoids is greatest when co-consumed with the source of the carotenoid. Interference with absorption of carotenoids diminishes and then disappears as the time between eating an olestra-containing food and a carotenoid-containing product increases.

The precise length of time olestra interferes with absorption varies with the dose of olestra, and also varies somewhat from individual to individual, as GI transit time is variable among individuals (61 FR 3118 at 3144).

In the clinical cross-sectional sample, 217 of 947 individuals reported eating at least one olestra-containing food in the previous month with a median intake of 8.1 g of olestra per month. The 90th percentile consumption level was 64 g of olestra per month. Of the 402 clinical cohort participants who were considered consumers of olestra, only 139 reported eating any olestra-containing foods in the previous month. The median frequency of eating olestra-containing foods for this group of consumers was 1.01 times per month with a median intake of 11.9 g of olestra per month. The 90th percentile

frequency of eating olestra-containing foods was six times per month with a 90th percentile intake total of 70.6 g of olestra per month.

FDA notes the infrequent and small olestra ingestion reported in the study. These reports are drawn from participants' “real-life” use of snacks made with olestra. FDA evaluated whether there were changes in serum levels of carotenoids and fat-soluble vitamins from year 0 to year 1 in the clinical cohort. FDA also evaluated whether olestra consumption was associated with changes in serum carotenoid status and fat-soluble vitamins. FDA noted the various changes in serum measures (a drop in total serum carotenoids as well as in concentrations of vitamins A and D, lycopene, lutein, zeaxanthin, and an increase in beta-cryptoxanthin) seen in the clinical cohort group from year 0 to year 1. FDA also noted that there is a lack of association in the clinical cohort between olestra ingestion and any nutrient changes, and therefore, the changes are unlikely to be caused by olestra consumption (Ref. 24).

2. Comments Regarding the Active Surveillance Study

FDA received comments about the Active Surveillance Study. FDA considered these comments and responds in the following paragraphs.

(Comment 9) Comments from CSPI and academia to the current petition asserted that P&G's Active Surveillance Study showing that olestra consumption produced no change in carotenoid levels provides little useful data because the subjects consumed only small amounts of olestra. CSPI stated that study subjects consumed no more than 2 g of olestra/day (approximately one-fourth to one-fifth of a serving of an olestra-containing snack per day) and only about 15 percent of adults in the study ate at least one olestra-containing snack per month. The comment from academia stated that any assumption about the effects of olestra on blood carotenoid levels should be based on the strong likelihood that at least some individuals will consume 1 to 4 oz of olestra-containing potato chips on a daily basis, the effects of which are addressed in the preapproval studies.

The Active Surveillance Study is only one piece of information in the current petition. It was designed to assess the effects of olestra consumption on serum carotenoids and fat-soluble vitamins under customary or usual consumption conditions. As such, it complements the preapproval studies, which were conducted using consumption scenarios designed to assess the safety of olestra's effects on serum carotenoids and fat-soluble vitamins. The highest dose of olestra consumed in the preapproval studies was 32 g/d, which is equivalent to eating approximately 4 oz of olestra-containing chips; in contrast to the Active Surveillance Study, in the preapproval studies olestra was consumed in a variety of foods for which it is not approved for use. FDA noted in the 1996 final rule that it was likely that olestra's effects on carotenoid absorption would be substantially less than those observed in the 8-week studies (61 FR 3118 at 3149). Under the conditions of the 8-week studies, which were designed to assess safety, FDA found supplementing olestra with vitamin A to compensate for the provitamin A function of beta-carotene addressed the possible safety concerns about carotenoid loss in olestra-containing foods. The comments provide no evidence to contradict FDA's 1996 conclusions.

FDA agrees that P&G's active surveillance did not identify high levels of olestra consumption. Importantly, however, the levels of olestra consumption identified in P&G's Active Surveillance Study provide information about customary or usual consumption which is relevant to the labeling issue raised by this petition.

E. Consultations and Literature Review Regarding Nutritional Effects

FDA considered data and information that became available after the 1996 decision in assessing whether the scientific understanding of the possible human health benefits of carotenoids has changed since FDA's 1996 decision, and whether new information should be reflected in the label statement.

The petitioner conducted a literature review of all peer reviewed articles published between January 1996 (when the 1996 final rule was published) and May 1998, just prior to the FAC meeting, concerning possible health effects of carotenoids. This review included more than 200 references to carotenoids and their possible role in human health (Refs. 25 and 26). The petitioner's conclusion was that the reviewed data did not establish that consumption of carotenoids confers protection from disease.

FDA considered data and information discussed at the 1998 FAC meeting. The petitioner presented its review of the scientific literature on carotenoids and human health. The petitioner sponsored a study that was presented to the FAC that found no significant association between macular pigment density with olestra intake. The researchers testified that the relationship between the carotenoid-rich macular pigment and the disease process was yet to be understood.

Transcript, vol. 2, pp. 197, 201, 210.

At the FAC, the petitioner called upon Dr. Gilbert Omenn

to present results from intervention studies with beta-carotene.

These studies indicated that there was an association between beta-carotene intake and increased risk for lung cancer within the study groups.

At the time Dr. Omenn was Executive VP Medical Affairs and CEO, University of Michigan Health System and was a principal investigator for the CARET study.

Transcript, vol. 2, pp. 154-160.

During the open public hearing portion of the FAC meeting, a number of individuals that the petitioner invited as experts spoke about the potential role carotenoids play in human health and expressed the view that carotenoids do not explain the cancer preventive effect of fruits and vegetables.

Transcript, vol. 2, pp. 32-34, 42-44.

At the FAC meeting, CSPI, and the individuals they called upon as experts, asserted that a consensus had been developing among the scientific community that carotenoids are likely to reduce the risk of certain chronic diseases. For example, Dr. Graham Colditz of Harvard Medical School said that a low intake of carotenoids is associated with an increased risk of cardiovascular disease and certain cancers.

Transcript, vol. 2, pp. 233-247.

Most members of the FAC expressed the view that epidemiological data show a decreased risk for certain chronic diseases and cancer with increased intake of fruits and vegetables. The increased intake of fruits and vegetables is associated with an increased serum level of carotenoids (which are a component of fruits and vegetables), but it is yet to be determined what, if any, specific role carotenoids play, and at what level they may be required in the diet.

Transcript, vol. 3, pp. 102-174.

The Panel on Dietary Antioxidants and Related Compounds, Food and Nutrition Board, Institute of Medicine (IOM), National Academy of Sciences (NAS) published a report in 2000 (Ref. 27). The panel that produced the report considered dietary antioxidants and other compounds to assess the required daily intakes for these nutrients. The NAS panel noted that there is a considerable body of research relating blood levels of carotenoids with a lower risk for some chronic diseases. However, the NAS panel concluded that this evidence did not support a requirement for carotenoid intake because the observed effects may be due

to other factors related to fruit and vegetable intake. Intervention studies designed to test whether carotenoids (specifically beta-carotene) had any direct protective benefits for health did not show any benefit compared to the control (placebo supplement), and indicated that there was an increased incidence in disease (lung cancer) for certain at-risk sub-populations (smokers). The panel did not propose establishing a dietary reference intake (DRI) for beta-carotene or any other carotenoid (Ref. 27).

FDA considered the NAS report on carotenoids and concluded that the evidence concerning carotenoids and the conclusions that could be drawn from the evidence about carotenoids and human health had not substantially changed since the 1996 decision. FDA acknowledges that investigations are continuing on carotenoids to better understand biochemical mechanisms and genetic controls of these substances, and what, if any, role carotenoids have in human health (Ref. 28).

In the fall of 2000, FDA consulted with the NEI for an update as to whether there had been a change in the understanding of the science regarding lipophilic carotenoids and eye health since FDA last consulted with NEI on this question prior to the 1996 final rule (Ref. 29). The NEI said that no specific vitamin or carotenoid had been established as protective against macular degeneration (Ref. 30). The NEI also said that the ongoing “Age-Related Eye Disease Study” (AREDS) includes a randomized clinical trial of an antioxidant combination (beta-carotene, Vitamins C and E) or zinc that is evaluating the effect of these nutrients on macular degeneration and cataracts.

Other investigations continue to explore the hypothesis that oxidative damage to the retina increases the risk of macular degeneration and that antioxidant nutrients and carotenoid pigments concentrated in the macula may protect against this damage (Ref. 30).

Since FDA's consultation with the NEI in the fall of 2000, the ongoing AREDS study published a report of a randomized clinical trial of an antioxidant combination (beta-carotene, Vitamins C and E) or zinc evaluating the effect of these nutrients on macular degeneration and cataracts (Ref. 31). Results of the study showed that a combination of antioxidants (vitamin C, vitamin E, beta carotene) and zinc reduced the probability for the development of advanced age-related macular degeneration (AMD) in study subjects who were at high risk for developing AMD. The groups given only antioxidants, or only zinc, did not show this reduction in rates of at least moderate visual acuity loss.

F. Consumer Perception Studies of the Label Statement

P&G and Frito-Lay submitted data from studies designed to test consumer understanding of the label statement required by the 1996 final rule as comments to that final rule. Additional reports of testing, conducted after the original comment period for the label statement closed on April 1, 1996 (61 FR 3118 at 3160), were also submitted to the agency. These reports are discussed in the following paragraphs.

1. 1996 Consumer Studies

On April 1, 1996, P&G submitted consumer studies conducted on the label statement required on olestra-containing foods. These studies were completed before olestra-containing foods were available in the marketplace. The petitioner did both qualitative (focus group) and quantitative (mall intercept, detailed questionnaire) testing. The objective of the qualitative research was to determine how consumers comprehended the required label statement and to develop potentially more informative label statement(s) for use in subsequent quantitative research. The objective of the quantitative research was to understand how the required label statement and alternative label statements communicate to consumers, to understand issues raised by the various label statements, and to understand how the label statements affect consumers' understanding of olestra-containing snack foods.

In the qualitative research, three focus group sessions were conducted in each of three cities (total of nine focus group sessions) among adults or teens. Participants saw a realistic product package and several possible versions of the olestra label statement, and were told that a version of these statements might appear on product packages. Participants discussed their impressions of the product and the various label statements in their own words. The group went through each label statement line by line.

In the quantitative research, a detailed questionnaire was presented to 1,726 respondents, adults and teens, recruited at shopping malls at 40 different sites around the country. Respondents were randomly assigned to a group to assess one of four conditions for wording or presentation of the label statement. Respondents were shown the assigned information statement on a realistic product package that included a nutrition facts panel, ingredient list, and other product information. They then answered questions about the product, the information statement, and the effects of olestra.

The petitioner concluded from these studies that the required label statement did not communicate clear and understandable messages to consumers. The petitioner found that most participants in the studies were confused by label statements about both the GI effects and the nutrient effects.

The petitioner asserted that the data demonstrate that consumers, after reading the “vitamins added” portion of the required label statement, are left with the impression that eating olestra-containing foods will change their vitamin status. After reading the nutrient statements, some participants inappropriately concluded that olestra is not safe based on presumed vitamin effects. The petitioner stated that the qualitative research indicated that when participants understood that there are no net consequences on vitamins A, D, E, and K, the participants questioned the need for any statement or were suspicious of the statement. The petitioner stated that this study shows that consumers find the concept of nutritional effects and compensatory addition difficult to comprehend without extensive amounts of information. The petitioner concluded from the results of the quantitative studies that a simple label statement indicating that the vitamins in the ingredient statement do not provide a nutritionally significant source best communicates to consumers the fact that there would be no effect on their status of vitamins A, D, E, and K.

Also, the petitioner concluded that the term, “other nutrients,” appears to provide no meaningful information to consumers. The petitioner reported that a majority of participants concluded that there were no effects on other nutrients regardless of whether the label statement cited effects on “other nutrients.” For those participants who did notice the message, they incorrectly concluded that a variety of nutrients, some known not to be affected by olestra (for example, vitamins C and B) were, in fact, being affected.

The petitioner stated that results from the focus group study on the GI portion of the label statement showed that the currently required label statement may cause consumers to incorrectly attribute GI symptoms to the consumption of olestra, including GI symptoms that olestra does not cause and GI symptoms that are not listed on the label statement. The petitioner said that the research supports the conclusion that the label statement may cause consumers to wrongly attribute symptoms because participants

interpreted the label statement in the context of their experience with other foods that are not labeled. Because there are other foods that cause GI symptoms but are not labeled (

e.g.

, psyllium,

wheat fiber, and beans), consumers infer from the olestra label statement that olestra's effects must be worse. The petitioner characterized a typical participant reaction during the focus group testing to be, “If it's like my other experiences, then why does it have this label?”

FDA assumes that the term “psyllium” refers to the soluble fiber component of the psyllium husk that is the subject of the agency's regulation in part 101 (21 CFR part 101) authorizing a health claim for soluble fiber from certain foods and coronary heart disease (§ 101.81). FDA considers both “psyllium seed husk” and “psyllium husk” to be common or usual names for this substance, but uses the term “psyllium” where it was used by the petitioner or comments.

The petitioner found that a label statement with an explanation of why olestra might cause GI effects (“Because olestra is not digested, it may cause intestinal discomfort or a laxative effect.”) added significantly to participant understanding. When specific symptoms were mentioned, such as “loose stools” and “abdominal cramping,” more participants responded that they would expect those GI symptoms, compared to panelists who viewed statements that did not mention specific symptoms. The petitioner also found that general GI symptom terms, such as “laxative effect” and “intestinal discomfort” communicate the same expectations in GI changes as the specific terms for other GI symptoms, especially for the range of symptoms related to stool changes.

The petitioner also investigated consumer reaction to the boxed configuration of the label statement, and concluded that statements not boxed had less connotation of harm.

Frito-Lay, an interested party and producer of olestra-containing snack foods, also submitted to the agency results from consumer studies on the label statement conducted prior to marketing of olestra-containing foods (sent as comments to Docket No. 87F-0179, dated March 28, 1996, and March 29, 1996). The purpose of these studies was to test the effect and communication value of the required label statement and alternative statements developed by Frito-Lay. The same type of methodology described above for the quantitative assay was used to obtain responses from 1,183 individuals from 5 sites around the country. Respondents were shown a label statement and then asked, based on this label statement, whether they believed that products containing olestra were safe. Frito-Lay said that in response to all the tested label statements, including the required statement, most respondents were uncertain as to the safety of olestra (66 to 71 percent), or thought it unsafe (14 to 19 percent). Because none of the label statements Frito-Lay tested eliminated consumer misconception about safety (including a label statement declaring that olestra has been found safe for consumption by the FDA), Frito-Lay concluded that there should be no special label statement. Additionally, Frito-Lay found that 63 to 65 percent of respondents believed that some people would experience GI discomfort. About half of these respondents said that they would delay going to a doctor if they ate a product containing olestra and then experienced GI discomfort for which they would normally seek medical attention. Additionally, a majority of the respondents (68 to 71 percent) believed that olestra would decrease the level of vitamins A, D, E, and K in their bodies, and a majority believed that other nutrients are affected by olestra.

FDA reviewed the consumer perception studies submitted in 1996 by P&G and Frito-Lay (Ref. 32). FDA noted that the studies were an attempt to evaluate what the olestra label statement communicated to consumers regarding several issues. These issues include the following subjects: (1) The safety of olestra; (2) whether the portion of the label statement about GI effects communicates reasonable expectations about the severity, frequency and duration of potential symptoms, and whether alternate wording or presentations communicate more effectively; and (3) whether the portion of the label statement about the potential nutrient absorption effects of olestra effectively communicates the reason for the addition of vitamins A, D, E, and K, as well as the scope and potential severity of the consequences of eating olestra, and whether alternate wording or presentations communicate more effectively.

FDA found the mall intercept studies to be adequate in methodology and sample size to differentiate between the communication effectiveness of the statements tested, including such changes as alternate wordings, or separation of portions of the label statement. For example, in part of Frito-Lay's quantitative study, participants were asked about the safety of olestra before, as well as after, viewing the test label statement. This use of a question before viewing the label statement serves to measure the impact of the label statement on participants' opinions or whether that opinion was established prior to viewing the label statement. However, the studies were limited to some extent by the choice and wording of questions. For example, P&G's quantitative study did not include, as a control, a “no information” statement, so the communication value of simply having a statement on the product package cannot be evaluated.

Regarding the safety of olestra, FDA found that the results of the consumer perception studies conducted by the petitioner and by Frito-Lay show that the label statement is misunderstood by respondents and thought to be a warning about possible health consequences of olestra consumption. FDA notes that Frito-Lay's data demonstrate that there was an increase in the level of concern about the safety of olestra after participants read a label statement. Specific wordings or presentations contributed little to the level of expressed concern. Only when a label statement included wording that FDA has found olestra to be safe for consumption was some of the concern alleviated. FDA also noted that when participants were given the opportunity to respond to the question of whether olestra is safe by opting for “uncertain,” a majority chose this response to every label statement examined. When response options were limited to yes or no, the majority chose “no” (

i.e.

, not safe).

Regarding GI symptoms, FDA concluded that there was no indication from these studies that consumer expectations about the severity, frequency, or duration of GI symptoms was influenced by specific wording or qualifications on effects or by whether there were any directions about when to contact a physician. There also was no indication from these studies that consumers' expectations about the GI symptoms were influenced by whether the nutrient portion of the label statement was present. Participants tended to use the same words used in the label statement to describe potential symptoms (

i.e.

, loose stools, abdominal cramping), but alternate words to describe certain GI effects (loose stools, more frequent bowel movements, diarrhea) were all understood to mean the same thing. When asked what proportion of the population might experience symptoms, modifiers had little effect on respondents' answers.

Regarding the nutrient absorption portion of the label statement, FDA found that the results of the 1996 consumer perception studies show that the current label statement is not effective in explaining the rationale for and quantitative consequences of adding the four fat-soluble vitamins to

olestra-containing foods. Respondents' knowledge about olestra's ability to interfere with the absorption of fat-soluble components of the diet was not tested directly, so it is not possible to assess the role that prior knowledge has in respondents' interpretation of the label statement.

The studies demonstrate that consumers do not understand that the addition of the vitamins was intended to produce no

net

effect in the body. The studies also show that respondents tended to believe that the statement about the inhibition of absorption applied to many other nutrients, including those on which olestra has no effect.

Without the absorption statement, somewhat fewer respondents believed that vitamin A, D, E, and K levels would be changed by consuming olestra, but fewer respondents were also aware that olestra reduced the absorption of these vitamins. Even without an absorption statement, substantial fractions of respondents believed that consuming olestra-containing foods would change both fat-soluble and fat-insoluble vitamin levels, presumably because of prior beliefs about olestra. Variations on the wording of the portion of the label statement regarding vitamin absorption and addition made consumers more aware of the vitamin absorption effect of olestra, but none remedied the miscommunication.

FDA concludes, based on this work, that neither the current label statement nor the alternative label statements tested on product packaging clearly communicate to consumers the effect of olestra on vitamin absorption. Without more detailed information or familiarity with olestra, consumers drew inappropriate inferences about the scope and magnitude of the additive's effect on vitamin absorption.

2. 1999 Consumer Studies

In 1998 and 1999, the petitioner conducted quantitative consumer research using a detailed questionnaire, and submitted the study to the agency on April 22, 1999 (Docket No. 00F-0792). The petitioner's stated purpose for this study was to obtain quantitative data on consumer perceptions of the required label statement. Participants were asked to respond to a series of questions regarding safety and GI effects after reading the required label statement.

The petitioner reported that 61 percent of participants thought that the products bearing the required label statement were unsafe.

The same percentage believed that the label statement was the government's way of telling them that the product was unsafe. A majority viewed the label statement as a warning, and not as an information statement. After reading the label statement, 83 percent of respondents believed that they could experience symptoms after eating a handful of chips, and approximately a quarter of these respondents would attribute extremely serious symptoms to olestra (severe diarrhea lasting several days, bloody stools, or vomiting lasting up to several days). The petitioner noted that extensive clinical data on olestra show that the additive does not cause such symptoms. The petitioner concluded from the results of this study that the label statement is misleading and conveys messages to consumers that are not consistent with the total body of clinical data on olestra or with FDA's intention in requiring the label statement.

By comparison, in the 1996 survey 45 percent of respondents, after viewing the required label statement, considered products containing olestra to be unsafe (Ref. 33).

P&G also sought to assess consumer perceptions about the current and alternative label statements. Twenty alternative label statements were tested and rated on a scale of 1 to 9 for the degree of safety perceived from the label statement (1 is “not at all safe,” 9 is “very safe”). For the GI portion of the label statement, P&G reported that respondents' perception of the degree of safety of olestra-containing foods after viewing the alternative GI statements ranged from 3.8 to 6.8. Alternative GI statements that provided a familiar frame of reference (comparison to beans, or onions for example), or that stated that GI symptoms were not a likely consequence, resulted in a greater perception of safety than those statements that provided generalized GI symptom data or context to qualify or describe GI symptoms. In this study, P&G found that the statements that elicited the lowest perception of safety were statements that specified GI symptoms, including those that are in the current label statement. For the nutrient portion of the label statement, P&G reported that more than 80 percent of study participants who read an ingredient declaration statement in which the vitamins A, D, E, and K were marked with an asterisk and accompanied with an explanatory phrase (“not a nutritionally significant source”) and that no longer had the phrase “other nutrients,” believed that levels of vitamins A, D, E, and K and other nutrients would not change after eating olestra-containing foods.

P&G also conducted research by a national tracking survey to measure consumer awareness of olestra and to determine whether consumers had concerns about olestra's potential GI effects (Docket No. 00F-0792). Survey results were obtained between January 1998 (just prior to the start of national marketing of olestra-containing foods) and May 1999. P&G reported that the results of the tracking survey showed that population awareness of olestra-containing foods increased substantially during this period (from 38 percent to well over 70 percent). The study also showed that respondents who were familiar with olestra-containing foods were quite concerned about possible GI effects. The percentage of “aware” consumers who were at least somewhat concerned about GI effects averaged 74 percent at the beginning of the tracking survey, and 70 percent after national marketing of olestra-containing food was fully underway.

Frito-Lay also conducted new studies on consumer perceptions of the olestra label statement to determine whether that statement was still capable of influencing consumer perception, as it did in 1996 (FAP 0A4708, exhibit 7 and August 13, 1999, Docket No. 00F-0792). Because the 1996 perception study was conducted shortly after FDA's approval of olestra but before availability of olestra in any product on the market, no participant in the 1996 study had eaten a product made with olestra. Frito-Lay therefore considers that the 1996 study showed the effect of the label statement on a “naive” population. Since the 1996 study, Frito-Lay points to numerous significant events involving olestra, including the nationwide availability of Frito-Lay products made with olestra, the FAC meeting held by FDA in 1998, and many national and local news stories about olestra. The new testing of the label statement was conducted with the protocol used in Frito-Lay's 1996 studies.

Frito-Lay reported that in its 1999 study, before seeing the required label statement, 64 percent of respondents were uncertain about safety, but only 6 percent said products made with olestra were unsafe. After viewing the label statement, the number of respondents who thought olestra products were unsafe more than doubled. No one who originally thought olestra products were unsafe changed their opinion after viewing the label statement. Frito-Lay presented results showing that only 24 percent of study participants concluded that products made with olestra do not affect the levels of vitamins in the body,

and an approximately equal distribution of participants concluded that olestra did or did not affect the absorption of other nutrients.

Frito-Lay concluded from these studies that the olestra label statement did influence consumer perception, much like it did in 1996. Frito-Lay also concluded that consumers still did not understand the various parts of the label statement and viewed it incorrectly as a warning.

FDA reviewed the consumer perception studies of the label statement submitted by P&G and Frito-Lay in 1999 and found them to be similar to the 1996 studies in methodology and types of questions asked (Ref. 33). This set of studies concentrated on the GI portion of the required label statement and perceptions of safety about products made with olestra. As with the 1996 studies, the studies were limited by choice and wording of questions and did not include, as a control, a “no information” statement. FDA notes that it is difficult, without careful controls, to distinguish whether a label statement miscommunicates information, is ignored, or is ineffective. Another important limitation to this study is the lack of measurement of initial attitudes toward olestra. Testing of initial attitudes and preconceptions is needed in order to identify the direction of the label statement's effect (

i.e.

, whether the consumer is more accurately informed or not informed).

FDA concludes that P&G's tracking study showed that consumers became more aware of olestra-containing snack foods as products were introduced nationally, and that increasing awareness was accompanied by concern about possible GI effects caused by these products. However, FDA also concludes that the tracking survey does not establish the role (if any) the required label statement plays in consumers' association of olestra-containing foods and GI effects. It cannot be determined from this study whether the rise in product awareness and association with GI effects was due to news reports, advertisements, promotions, or reading the label statement.

FDA concluded that the 1998 and 1999 studies reinforced conclusions from the 1996 studies and support several new conclusions. The new conclusions include the following scenarios: (1) Consumers became more concerned about the safety of olestra-containing products between 1996 and 1998, prior to the introduction of olestra-containing food into national distribution; (2) consumers familiar with olestra-containing snack food (Olean brand name) are very likely to make an association between olestra and possible GI effects; and (3) consumers newly introduced to olestra-containing products by the introduction of Olean brand products into local stores and the accompanying advertising and promotion are just as likely to make the association between olestra and GI effects as those who already knew about Olean products.

FDA notes that the 1998 study gave respondents a choice of “I don't know” about safety. Given this option, 45 to 49 percent of participants in the 1998 study chose the response “I don't know” when asked about safety, and only 10 to 13 percent of participants in the 1998 study chose the response “unsafe.” In contrast, in the 1996 survey, when participants did not have an option to choose “I don't know” when asked about safety, 38 to 61 percent of participants chose the response “unsafe.”

G. 1998 FAC Discussion of the Label Statement

The FAC discussed the required label statement on the last day of the 3-day meeting (June 17, 1998), after new data and information concerning possible GI and nutritional effects were presented and discussed. The FAC was to consider whether, in light of the new data and information concerning the consumption of olestra, the label of olestra-containing products should be changed in any way. The FAC was also asked to consider what factual information, if any, regarding the consequences of consuming olestra-containing products should be disclosed on the product label. FDA began the session by discussing the scope of the agency's authority, under the act, regarding labeling. The sponsor (P&G) made presentations to the FAC, followed by Frito-Lay, and CSPI. The FAC asked questions at the end of each presentation. Following this portion of the meeting, each FAC member was polled regarding the label statement.

Polling the individual members of the FAC about whether the label statement should be changed revealed a wide variation in opinions on the labeling issue.

A majority of members, however, did agree that the label statement should be modified in some way in order to make its messages more clear to the consumer. Several members stated that some labeling information about olestra was needed, based in part on their view that olestra-containing foods were still relatively new products and that consumers were not entirely familiar with these products. Some members of the FAC suggested there be a sunset clause on the label statement because after consumers became familiar with olestra, there would no longer be a need for a label statement.

Transcript, vol. 3, pp. 101-174.

Members made various suggestions for different wordings of the label statement to clarify to consumers the likelihood that olestra would cause GI effects and the nature of those effects. Other members expressed concern that consumers might confuse olestra's effect with more serious GI symptoms. Some members of the FAC concluded that olestra's GI effects did not warrant a special label statement, especially because consumers might mistakenly attribute more serious GI symptoms to the olestra. Other members thought a label statement should include information to tell the consumer to seek medical attention if symptoms persist. Several members said they believed that the current data did not support keeping the portion of the label statement on abdominal cramps.

A majority of the members of the FAC specifically agreed that the portion of the label statement regarding “vitamins added” should be removed and replaced with an asterisk following the vitamins in the ingredient listing and a footnote indicating that the added vitamins are not a nutritionally significant source or not nutritionally available. Both P&G and CSPI agreed that this approach was an acceptable and effective way to explain that the presence of the vitamins in the ingredient listing is not meant to imply that these foods are a source of these vitamins.

Transcript, vol. 3, pp. 92-93.

A majority of members of the FAC agreed that there is no basis, at this time, for adding back any carotenoid, and that the role carotenoids play, if any, in human health is not yet understood. Members did say that the sponsor and the agency should be aware of the evolving understanding of the health effects of carotenoids, and consider that information and its bearing on the use of olestra. Some members expressed reservations about not having any statement on the label about olestra's potential to interfere with the absorption of fat-soluble nutrients such as carotenoids, and suggested that the statement about “other nutrients,” might be clarified by changing the phrase to “nutrients found in fruits and vegetables.”

IV. FDA's Conclusions

A. The Applicable Legal Standard

Under section 409(c)(3) of the act, a food additive shall not be approved if

such approval would result in the misbranding of a food containing the food additive. Misbranding includes labeling that is misleading because it fails to reveal facts material with respect to consequences resulting from use of the additive under “customary or usual” conditions (sections 201(n) and 403(a)(1) of the act). Thus, the data and information of principal relevance to evaluating whether olestra must bear a label that discloses, for example, the possible GI effects of olestra, are those that evaluate the additive's effects when eaten at levels, and in patterns of consumption, that are customary or usual.

As the preceding discussion reflects, FDA has considered all of the evidence in the record and has considered the preapproval studies in light of the postapproval investigations reflecting customary or usual patterns of consumption.

This would be in contrast to the preapproval studies which were designed to assess safety.

In addition, it is critical to recognize that the issue presented by this petition is not whether olestra is safe for use in savory snacks; that issue was addressed in the 1996 final rule. Instead, the question before the agency is what labeling, if any, must be required for foods containing olestra to ensure that they are not misbranded (section 403(a)(1) of the act). Accordingly, the act's “reasonable certainty of no harm” safety standard in § 170.3(i) (21 CFR 170.3(i)), does not apply here.

FDA does not ordinarily require special labeling on a food that may have consequences of consumption (such as GI effects) when the available information shows that consumers are aware that such food cause the effects. Psyllium husk is an example of an ingredient that may cause GI effects. Consumers are aware of these p

This text is long and has been trimmed here. Open the source document for the complete record.

This is a copy of a public record, reproduced as it was published. It is not legal advice, and it may not be the version a court would rely on. Check the official source before you cite it.