Olestra is a mixture of sucrose polyesters formed from the addition of fatty acids to six, seven or eight of the hydroxyl positions on a sucrose backbone. Saturated and unsaturated fatty acids with a chain length ranging from C₁₂ to C₂₀ and higher can be used in the manufacture of olestra if the final product meets established fatty acid compositional specifications and the fatty acids are derived from edible sources (Federal Register 1996). The fatty acids used to produce olestra impart the same physical properties, taste and cooking characteristics, and heat stability profiles to olestra as found in the starting triglyceride (Gardner and Sanders 1990, Henry et al. 1992, Kester 1993).

Olestra (Olean, Procter & Gamble, Cincinnati, OH) is approved as a replacement for up to 100% of the fats and oils used in the preparation of savory (i.e., salty or piquant but not sweet) snacks such as flavored and unflavored chips and crisps, flavored and unflavored extruded snacks, and crackers (Federal Register 1996).

The purpose of the dietary assessment study described here was to determine the potential estimated daily intake (EDI)³ of olestra from the consumption of savory snacks prepared with olestra and to determine how (i.e., how often, in what context) savory snacks are consumed. This was done by analyzing 14-d food diaries collected by Market Research Corporation of America (MRCA) Information Services (Des Plaines, IL) from 4741 individuals during 1991-92. The data were analyzed to provide estimates of the following: 1 ) potential chronic (14-d average) and acute (amount eaten in a single day) EDI of olestra from the consumption of savory snacks, 2 ) potential frequency at which olestra will be eaten as defined by the number of times and the number of days savory snacks are normally eaten over a 14-d period, and 3 ) potential pattern of olestra consumption as defined by main meal and in-between meal eating occasions for savory snacks within a 14-d period. The EDI were determined on a gram per person per day basis but for simplicity will be referred to in gram per day (g/d) throughout this paper.

The MRCA Menu Census method was chosen because of its large size (e.g., about 2000 households and 5000 individuals), its duration of assessment (14 consecutive days/subject), its overall time frame (i.e., the survey is conducted annually and runs continuously throughout a calendar year to capture seasonal changes in eating habits) and the similarity of the survey's demographics (e.g., geographic location, age, gender, race, income, education and household size) to U.S. census statistics.

METHODOLOGY AND RESULTS

The MRCA Menu Census survey has been described in detail by Abrams (1992) and will be described only briefly here. As an overview, the survey tracks all foods and beverages (except table salt, table pepper and tap water) consumed by individuals daily, at home and away, at main meals and in between meals throughout a consecutive 14-d period. Key elements of the survey are provided below.

Survey population. The food diaries were collected from a population sample of 4741 men, women and children, representing about 2000 households, throughout the year. The daily diaries are completed by homemakers who were also long-term members of MRCA's National Consumer Panel and Weekly Purchase Diary Panel and who were trained and experienced in reporting personal and family eating habits in great detail.

Households included in the survey were nationally representative according to such criteria as geographic location, household size and household income. An average of 5.5 new households started their biweekly reporting every day of the year, so that each day of the week and of the year was equally represented. The demographic characteristics of the households were comparable with U.S. census demographics (Table 1). A final questionnaire included the self-reported age, gender, pregnancy status, weight, height and diet status of each household member. Other household demographic characteristics are collected separately in an annual questionnaire. Compared with the U.S. population in general, the survey population contained an adequate representation of children and a representation of individuals 55 years of age greater than that found in the U.S. population (Table 2). These two subpopulations might be of greatest concern with respect to the effects of olestra on nutritional status.

Table 1. Demographic characteristics of households in the 1991-92 MRCA Menu Census survey and the U.S. census1 [View Table]

Table 2. Person characteristics of the households in the 1991-92 MRCA menu census survey and the U.S. census1 [View Table]

Data collected. The diaries included a detailed description of each dish eaten and items added to it at the time of preparation or at the time it was eaten, whether it was eaten at home or away from home, whether it was eaten at breakfast, lunch or dinner or consumed at a morning, afternoon, evening or bedtime snack eating occasion, the position of the dish in the meal (i.e., first, second or third course) and which household members ate the dish. Portion sizes were not reported directly by the homemakers, but rather were estimated from the 1987-88 Nationwide Food Consumption Survey (NFCS) as described below.

Number of times savory snacks were eaten. Determination of the frequency of savory snack consumption was based on a detailed list of all commercial products that could contain olestra as defined in MRCA's code book of over 20,000 commercial products. The list was continuously updated as new products came on the market. The type of savory snacks tracked in the survey included: curls, puffs and extruded snacks; corn chips; tortilla, nacho and taco chips; potato chips; baked crackers and chips; saltine and oyster crackers; butter- and other-flavored snack crackers; and filled and sandwich-type crackers.

Determination of average amount (i.e., portion size) of savory snacks consumed per eating occasion. The savory snack portion sizes used to estimate olestra EDI were computed by MRCA, for children by age groups, and for adults by age within sex groups, from the USDA/NFCS detailed eating records according to the weighted grams eaten and smoothing these averages over age groups with the use of a method developed under the direction of the FDA and The National Academy of Sciences Food Additives Review Committee. The 1987-88 NFCS (U.S. Department of Commerce 1988) study collected food eating data from each respondent for up to three consecutive days. The data for d 1 were obtained by a trained interviewer by the 24-h assisted recall method. The data for the next 2 d were self-reported by adults, and by the homemaker for children, in separate daily diaries, and collected by the interviewer on a following day. MRCA used only those respondents who reported their eatings for all 3 d, for whom weights were available to balance the sample to the demographic characteristics of the U.S. census and to the days of the week.

The portion sizes ranged from 11 g per person per eating occasion for 2 to 5-y-old children eating crackers to 64 g per person per eating occasion for 45- to 54-y-old males or females eating sandwich-type crackers (Table 3). The portion sizes for males were greater than those for females with one exception; females 18-24 y of age ate larger portion sizes of flavored crackers. The portions sizes for sandwich crackers were the same for males and females. MRCA did not separate children's portion sizes by gender.

Table 3. Range of 50th-percentile portion sizes of savory snacks based upon 3-d reporters in the USDA Nationwide Food Consumption Survey of 1987-88 [View Table]

The reasonableness of using the 1987-88 NFCS average portion sizes for savory snacks was assessed by analyzing savory snack portion size data from the 1994 Continuing Survey of Food Intake by Individuals (CSFII) (USDA 1996) and comparing those sizes with the NFCS sizes. The distribution of portion sizes reported by those respondents to the 1994 CSFII who reported eating chips and/or crackers on any of the two, widely separated, nonconsecutive days of that survey was examined. The CSFII data are from quantitative 24-h recall of food intake, probed for detail by trained interviewers. Chips were reported in the 1987-88 NFCS as two separate categories, namely, white potato chips, favored and unflavored, and a second category that included corn-based puffs and twists, tortilla chips and corn chips. The entire combined second category was used for examination of the 1994 CSFII data. Figure 1 shows the mean portion sizes for white potato chips (identified as Chips 1) and tortilla and corn chips and corn puffs (identified as Chips 2) for the 3-d average of the 1987-88 NFCS and for all chips on d 1 and d 2 of the CSFII survey for males and females. Figure 2 shows the same data for crackers.

The mean CSFII portion size of crackers was slightly greater than the NFCS portion size for the same snacks. There was no clear trend for chips. The serving sizes were similar for females; the CSFII serving sizes for teenage and young adult males tended to be greater than the NFCS sizes.

Determination of the potential olestra concentration in savory snacks. To define the potential olestra content of savory snacks, snack products representative of the different types and methods of manufacture in the industry were analyzed for total replaceable fat content. The percentage (wt/wt) of replaceable fat for each type of snack product was determined by using methods approved by the Association of Official Analytical Chemists (AOCS 1980). Duplicate samples were analyzed and the results were averaged for each type of product. The samples analyzed were selected to include the unflavored and flavored market leaders and additional products, as necessary, to represent the range of taste and texture within each type of savory snack (e.g., sliced potato chip or sandwich cracker).

The fat content of reduced-fat products was not analyzed but rather assumed to be the same as the content of their full-fat counterparts. In addition, any snack food captured in the MRCA census survey and not analyzed directly was assumed to have a replaceable fat content equal to the greatest measured value in that particular product category.

The replaceable fat content (wt/wt %) for the savory snack products assessed in the MRCA survey is summarized in Table 4. The replaceable fat content for potato chips, fabricated chips, corn and tortilla chips, and extruded puffs and curls averaged 35% with a range of 27-43%. The replaceable fat content for soda crackers, flavored and unflavored crackers, cracker chips and filled crackers averaged 19% with a range of 1-27%.

Table 4. Range of replaceable fat content of savory snacks as determined from analytical measurements [View Table]

Table 5. Estimated mean and 90th-percentile (90th %) intakes of olestra from the consumption of savory snacks, eaters only [View Table]

The estimated mean acute olestra intake for males and females combined ranged from 6.3 to 14.5 g/d, depending upon age. For the total population, all ages, the mean value was 10.2 g/d. For the 90th-percentile consumer, acute intakes for males and females combined ranged from 11.7 to 23.9 g/d, depending on age; for all ages combined the estimated intake was 18.3 g/d. Males and females in the age range 13-44 y had the largest intakes. Males had larger estimated intakes than females at each age group. The estimated mean and 90th-percentile acute olestra intakes for pregnant females 12-55 y of age were similar to the estimated intakes for nonpregnant females of the same age range. Dieters, both those attempting to control calorie intake (low calorie) and those attempting to control fat intake (low fat), had slightly lower estimated intakes than the average nondieter.

The estimated chronic (14-d average) intakes were less than the estimated acute intakes, as might be expected from the inclusion of their noneating days in their averages. For males and females combined, the mean intakes ranged from 2.0 to 4.4 g/d, depending on age. The mean value for the total population (all ages) was 3.1 g/d. The estimated 90th-percentile chronic intakes for males and females combined ranged from 4.6 to 10.0 g/d, depending on age; the value was 6.9 g/d for all ages combined.

As with the acute intakes, the largest chronic intakes were found for 13- to 44-y-old individuals, either male or female, with males having larger intakes than females at each age group. Pregnant females had estimated chronic intakes somewhat lower than those of nonpregnant females of the same age range, and dieters tended to have slightly lower intakes than the average nondieter.

The number of times savory snacks were consumed out of the 14-d survey period is presented as a histogram in Figure 3. Among savory snack eaters only (all ages, both genders), savory snacks were eaten four times in 14 d by the 50th-percentile consumer and about five times by the average consumer. The 90th-percentile consumer ate savory snacks 10 times in 14 d. The maximum number of eatings in 14 d was 35. 

The number of days savory snacks were eaten during the 14-d survey period (all ages, both genders) is shown in Figure 4. Savory snacks were eaten 3 d out of the 14 by the 50th-percentile consumer. This value increased to 8 d for the 90th-percentile consumer. The maximum number of days was 14 of 14. 

On average, 69% of savory snack eatings were with main meals, whereas the remainder were consumed as in-between meal snacks. For the 50th-percentile consumer, only 7% of main meals included savory snack products and for the average consumer, only 8% of main meals. At the 90th percentile, savory snacks were included with 18% of main meals.

DISCUSSION

This methodology was also used to assess olestra eating patterns in the context of frequency of consumption at main meals. Although savory snacks are eaten on average about 69% of the time with main meals, these snack foods are eaten infrequently in the overall context of the total diet. The 50th-percentile snack consumer eats savory snacks with only 7% of main meals across a 14-d period; the 90th-percentile consumer eats them with 18% of main meals in 14 d.

The MRCA Menu Census methodology provides a realistic estimate of the consumption of savory snacks, and the intake estimates are sufficiently conservative to be useful in nutritional safety evaluation. The method is realistic because the survey relies upon the determination of the frequency of food consumption over 14 continuous days for each survey participant. In contrast, other surveys such as the NFCS or the CSFII evaluate food intake over a period of only 1-3 d (Anderson 1988). Because of the longer observation period, the MRCA method provides a relatively more representative long-term perspective on eating patterns and therefore is well suited to the estimation of chronic intake levels of dietary components such as olestra. The MRCA consumption data are collected yearly, providing an up-to-date resource of food consumption patterns that are inherently dynamic and were current at the time of this assessment, 1993. In addition, the survey runs continuously throughout the year to capture seasonal changes in eating habits. Data are also recorded for specific brands and types of foods such as savory snacks, which maximizes the accuracy and specificity of the intake estimates. Further, new brands and types of snack foods are continuously added to the survey as they come onto the market.

The methodology used in this assessment provides conservative intake estimates because the MRCA model assumes that 100% of all savory snacks available in the marketplace, including stores, restaurants or vending machines, are replaced by olestra-containing savory snacks and thus yields intake estimates much greater than can be reasonably expected from introduction of a new product line into the savory snack food category. In addition, the derived intake estimates were adjusted upward by 10% to provide more conservative estimates. The degree of conservatism inherent in the method has been assessed by evaluating the intake of aspartame in Canada on the basis of a premarket "theoretical" model, which, like the MRCA Menu Census Survey, assumed that aspartame foods captured 100% of market share. For comparison, aspartame intake was also estimated on the basis of a postmarket "actual monitoring" program. The results of this analysis demonstrated that mean aspartame intake estimates derived in the postmarketing monitoring program were one tenth (10%) of the premarket, total market penetration model estimates (Butchko and Kotsonis 1991, Lauer and Kirkpatrick 1991).

Despite the strengths and conservatism of the MRCA methodology noted above, any method applied to estimating potential food or nutrient intake has inherent weaknesses that may, if significant in nature, undermine the ultimate value of the assessment outcome. Potential weaknesses of the MRCA survey are the use of the 1987-88 NFCS data to estimate portion sizes of savory snacks and the possible underreporting of savory snack consumption in the Menu Census survey. An assessment of each of these follows.

The MRCA Menu Census survey does not include quantitation of portion sizes; instead it relies on other appropriate and current data bases for that information. At the time this assessment was made, 1993, that data base was the 1987-88 NFCS. The 1987-88 NFCS has been criticized because of its low response rate relative to other USDA national food consumption surveys. For example, the response rate was 38% compared with a target of 74%, and lower than the 61% response rate for the 1977-78 NFCS. In addition, the response rate was variable over both time and geography. Another criticism of the 1987-88 NFCS is the lack of availability of data on the sociodemographic characteristics of the nonresponding households.

An analysis of the effects of the structure of household-level nonresponse has been conducted by Aickin et al. (1992) using selected intake variables. Briefly, these researchers used the 1977-78 NFCS as an external data base with the same sampling design and examined the relationship of area segment-level nonresponse to the differences in intake estimates and in their variability between the 1987-88 and 1977-78 surveys. The variables examined were as follows: 1 ) in each case separately for individuals < 19 y and 19 y of age, energy intake and the proportion of meat, poultry and fish intake, which was assumed to change throughout the population in the interval between the two surveys, and 2 ) for women 19-50 y of age and children 1-5 y of age, intake of the 10 food groups most commonly reported in the 1985 CSFII. For women, these were milk, eggs, white bread, cucumber, lettuce, onion, butter, margarine, salad dressings/mayonnaise, sugar, coffee, tea and soft drinks. For children, they were milk, bologna, sausage, eggs, white bread, potato, butter, margarine, sugar, soft drinks, orange juice and fruit-flavored drinks.

This analysis showed no consistent effects of the nonresponse structure (i.e., the attributes and demographics of the nonresponders) on intake. It would not be expected that the household-level nonresponse rates or the geographical or time-related structure of nonresponse would affect the estimation of the average amount of food consumed per eating occasion unless there was a systematic bias toward nonparticipation of households in which this average amount of food consumed was different than that of other households. Although it is not empirically possible to ascertain that this did not occur, it is unlikely that it did because subsequent population intake surveys (e.g., 1996 CSFII) have had good household response rates. This demonstrates that nonresponse is not a secular trend but rather an indication that, in the NFCS, the survey team may have failed to follow procedures designed to ensure a good response rate. Therefore, the portion size estimates from the 1987-88 NFCS reasonably represent portion sizes of the U.S. population.

Table 6. Total energy intake and the percentage from fat for the 1987-88 NFCS and the 1991-92 MRCA surveys and the percentage differences between the two data sets [View Table]

Examination of the distribution of portion sizes for savory snacks in the 1994 CSFII data shows a larger average portion size for crackers, but not for chips, relative to the 1987-88 NFCS data. Males 18-24 y of age, for example, averaged 22.1 g crackers/serving in the NFCS and 42.8 and 60.5 g on d 1 and 2 of CSFII, respectively. It is not possible to ascertain whether the difference is due to differential reporting accuracy, differences in the sample or a secular trend toward larger portion sizes of crackers between the two time periods. These differences in cracker portion sizes may also be due to the fact that personal interviewing concerning eating habits by trained individuals was done on each of the 2 d of the CSFII study. In contrast, this approach was used only on d 1 of the NFCS study, then followed by 2 d of self-reporting in diaries by the untrained respondents. The significance of these differences in estimating olestra intakes is likely to be minimal in light of the conservative assumptions applied in the analysis.

A common criticism of any population dietary survey is that the respondents may have bias for not accurately reporting the frequency of consumption of certain foods or beverages. This potential may be especially problematic for the MRCA survey given that underreporting could be characterized as an eating occasion that occurs without the knowledge of the head-of-household responsible for recording food consumption frequency for all household members. If the consumption of savory snacks was not accurately reported to the head-of-household by other household members, then olestra intake might be understimated.

As noted above, the MRCA survey method assumes that 100% of all savory snacks are replaced by olestra-containing savory snacks. This can lead to a significant (i.e., likely an order of magnitude) overestimation of potential olestra intake among savory snack consumers. Although this magnitude of conservatism would appear to negate any concern for underreporting of savory snack food intake, further perspective on this issue is provided below.

It is reasonable to assume that if MRCA survey participants underreport savory snack consumption, then they do so consistently. There is general recognition that people tend to overestimate, on either report or recall, intake of food items they believe to be good for them and, conversely, underestimate those they believe to be bad. However, MRCA estimates for both total energy intake and fat intake are generally within 5% of averages from the 1987-88 NFCS (Table 6). Reported energy intakes showed no consistent differences across age or gender. The percentage energy from fat was consistently higher in the MRCA survey but the differences were slight, from 0.6 to 2.1%, depending on age or gender.

An additional characteristic of the MRCA data is that the food intake data are collected as one component of a multidimensional consumer purchasing and behavior inventory. There is not a particular focus on food or on savory snacks, thus possibly decreasing any reporting bias that might be present if the main issue were food intake and the individual or organization requesting the intake information represented the health profession.

The assessment described here provides a conservative estimate of potential olestra intake from savory snacks. The methodology used to derive the estimated intakes was demographically balanced, representative of U.S. households and provided the best perspective available on long-term eating habits among U.S. consumers. The estimated intakes served as the basis for evaluating the safety of olestra for use as a replacement of fat in savory snacks and were used to choose appropriate intake levels for the human and pig nutrition studies described elsewhere in this issue. The overall objective of these studies was to assess the potential for olestra to interfere with the absorption and utilization of macronutrients and water- and fat-soluble micronutrients. Toward that end, it was paramount to have not only an estimate of potential olestra intake on a gram per day basis but also an understanding of how olestra will be consumed (e.g., the number of times likely to be eaten per day, the number of days per week olestra is likely to be eaten and the likely frequency of consumption with main meals).

ACKNOWLEDGMENTS