Patterson RE, Kristal AR, Peters JC, Neuhouser ML, Rock CL, Cheskin LJ, Neumark-Sztainer D, Thornquist MD. Changes in Diet, Weight, and Serum Lipid Levels Associated With Olestra Consumption. Arch Intern Med. 2000;160(17):2600-2604. doi:10.1001/archinte.160.17.2600
Specially manufactured low-fat and nonfat foods have become increasingly available over the past 2 decades and controversy has surrounded the issue of whether these products have beneficial or adverse effects on the health and nutritional status of Americans.
This study examines the association of olestra consumption with changes in dietary intakes of energy, fat, and cholesterol and changes in weight and serum lipid concentrations. Data are from a cohort of 335 participants in the Olestra Post-Marketing Surveillance Study sentinel site in Marion County (Indianapolis, Ind). Diet, weight, and serum lipid levels were assessed before the market release of olestra and 1 year later, after olestra-containing foods were widely available. Olestra intake at the 1-year follow-up was categorized as none, low (>0 to 0.4 g/d), moderate (0.4 to 2.0 g/d), and heavy (>2.0 g/d).
Participants in the heavy olestra consumption category significantly reduced dietary intake of percentage of energy from fat (2.7 percentage points, P for trend, .003) and saturated fat (1.1 percentage points, P for trend, .02). Consumers in the highest category of olestra consumption had statistically significantly reduced total serum cholesterol levels of –0.54 mmol/L (−21 mg/dL)compared with –0.14 mmol/L (−5 mg/dL) among olestra nonconsumers (P for trend, .03).
These results indicate that introduction of a new fat substitute (olestra) in the US market was associated with healthful changes in dietary fat intake and serum cholesterol concentrations among consumers who chose to consume olestra-containing foods.
THE LAST 10 to 15 years have seen an unprecedented increase in the availability of specially manufactured low-fat and nonfat foods1 and a reduction in the percentage of energy obtained from fat in the diets of Americans.2 This increase in nonfat and low-fat replacements for full-fat foods has also been paralleled by a notable increase in the prevalence of obesity in the United States such that almost 55% of adults are overweight or obese.3,4 Although there is clearly no evidence for cause and effect between these 2 events, their coincidence does highlight an important research question for public health nutrition policy regarding the effects of reduced-fat products on the health and nutritional status of Americans. Of particular interest is the effect of newly introduced foods made with the nonenergy fat substitute, olestra. Unlike many other fat-modified foods, which may be highly dense in energy, those made with olestra are substantially lower in energy than their full-fat counterparts.
To understand how olestra-containing foods can affect diet, it is useful to consider the ways olestra can alter dietary intake and the nutritional consequences of each pattern of use. Behaviorally, persons choosing to eat olestra-containing foods could add them to their diet in 1 of 3 nonexclusive ways: (1) by substituting olestra-containing foods for their full-fat counterparts (eg, eating olestra potato chips instead of full-fat chips), (2) by replacing other foods with olestra-containing foods (eg, eating olestra chips instead of fruit), or (3) by adding olestra-containing foods to the diet without making other changes. The nutritional consequences of these changes are primarily as follows: (1) if olestra-containing foods are substituted for the full-fat version, both energy and total fat intakes will decrease; (2) if olestra-containing products replace other foods, changes in energy, fat, and other nutrients will depend upon the foods that are replaced; and (3) if olestra-containing products are added, energy intake will increase but total dietary fat will not.
The effect of adding olestra-containing foods to the diet is complicated by physiologic and cognitive factors. There is considerable controversy about how dietary fat (and therefore fat substitutes) affects hunger, satiety, and the regulation of food intake. This question has been addressed extensively using experimental feeding studies in which the energy density of foods is covertly manipulated (ie, a defined preload is given) to observe subsequent effects on food intake in an ad libitum test meal. Much of this work has focused on olestra and has been previously summarized.5 Although the data are not entirely consistent, collectively, these studies suggest that consumption of reduced-fat foods is associated with somewhat reduced energy and/or dietary fat intake, which could lead to modest weight loss or prevention of weight gain.5 However, there are design aspects of experimental studies that limit our ability to draw conclusions about the effects of fat-modified foods on dietary intake in free-living adults. These limitations include the use of an artificial environment, small sample sizes, short study duration, and covert nature of the diet manipulation in most of these studies, which precludes any increased food consumption that might occur if the test subjects knew they were eating reduced-fat foods.6
Another potential effect of adding olestra-containing foods to the diet is that consumption of these nonfat foods could lead to unrestrained eating. One study of 95 males and females classified as unrestrained or restrained eaters specifically compared overt and covert substitution of full-fat chips with olestra-containing chips. When participants were unaware of whether they were consuming regular or fat-free olestra-containing potato chips, there was no evidence for a compensatory increase in energy or fat consumption during a 10-day period.7 However, when information on the fat and energy content of chips was provided, restrained eaters ate about 15% more fat-free chips than regular chips, although this increase in consumption did not eliminate the reductions in fat and energy intake associated with the fat-free chips. This study provides some evidence that Americans may use fat-modified foods as license to eat more of the same, or other, foods.5,8
None of the research to date addresses what happens to energy and fat intake among free-living adults who choose to consume fat-modified foods as part of a mixed diet. In this article we present results from the first year of the Olestra Post-Marketing Surveillance Study (OPMSS). Specifically, we examined changes in dietary intake and a variety of objective diet-related indicators (weight and serum lipids) among a cohort of adults assessed prior to the market introduction of olestra-containing savory snacks and about 1 year after these snacks were available to consumers.
The overall objective of the OPMSS is to assess olestra consumption and its association with serum concentrations of fat-soluble vitamins and carotenoids in population-based samples of US consumers. The OPMSS consists of 4 clinic sites, an operations support unit at Westat Inc (Rockville, Md), and a coordinating center at the Fred Hutchinson Cancer Research Center (Seattle, Wash).9 This study was approved by all institutional review boards and participant consent was obtained verbally (for telephone interviews) and in writing (for clinic interviews).
The sentinel data collection site was in Indianapolis, Ind, which was the location of the first major market introduction of olestra snacks. The OPMSS has several components, including a random-digit dial telephone survey, clinic cross-sectional samples, and a cohort. Data for this report are from the sentinel site and were collected at baseline, prior to the availability of olestra-containing foods in the local market, and at a second clinic visit approximately 1 year later, after the introduction of olestra-containing savory snacks. The analysis is confined to the cohort sample, which was chosen from the cross-sectional sample (n = 1962) interviewed at the clinic in Indianapolis between September 1996 and January 1997. The cohort was selected for higher levels of olestra consumption, based on dietary information collected from telephone interviews that were conducted approximately 3, 6, and 9 months after the baseline clinic visit and after olestra snacks appeared in the marketplace. Specifically, we recruited all adults who reported at least 2 occasions of eating olestra-containing foods, a random sample of 67% of those adults who reported a single eating occasion, and a random sample of 16% of those adults who did not report eating olestra. Of the 478 adults invited into the cohort, 88% completed the year-1 follow-up clinic visit. The final sample consists of adults with complete dietary and covariate data (n = 335) or complete weight, serum lipid, and covariate data (n = 326).
Nutrient intakes (dietary energy, fat, saturated fat, and cholesterol) were assessed using a food frequency questionnaire (FFQ)10 with a supplementary questionnaire on savory snack consumption. Participants were asked about dietary intake in the past month. The supplementary questionnaire assessed the intake of olestra-containing snack foods at year 1. The nutrient database (including olestra) was derived from the University of Minnesota Nutrition Coordinating Center nutrient database (Minneapolis).11 We excluded FFQs with fewer than 3350 kJ or greater than 20,920 kJ for men (n = 20) and fewer than 2510 kJ or greater than 16,740 kJ for women (n = 25) because these energy intakes suggest that participants did not complete the forms in a reliable manner.
Trained staff measured height and weight at the clinic visits and administered questionnaires on demographic characteristics (age, sex, race, education) and health-related behavior, such as physical activity. Fasting blood samples were drawn from all participants. Specimens were stored at −80°C until analysis.
Serum cholesterol and triglycerides were analyzed at Quintiles Laboratories (Atlanta, Ga) using enzymatic methods. Details on blood collection, processing, and analysis are given elsewhere.12 Precision and bias were less than 3%, as evaluated using packaged reagents, pooled human serum, and control serum.
Linear regression was used to model changes in dietary intake and diet-related factors by olestra consumption. We categorized olestra intake (the independent variable) into 4 levels based on mean intake (in grams per day) in the previous month: no intake (0 g/d), low intake (>0 to <0.4 g/d), moderate intake (0.4 to 2.0 g/d), and heavy intake (≥2.0 g/d).
The dependent variables were change in dietary intake, change in weight, and change in serum lipid concentrations. All models included age (continuous), sex, race (white and nonwhite), education (≤12, 13-15, ≥16 years), and baseline body mass index (calculated as weight in kilograms divided by the square of height in meters [kg/m2]). For each model of change in dietary intake, we included the baseline dietary intake being investigated. For example, for the model of change in percentage of energy from fat, baseline percentage energy from fat was included. For all models of change in weight and serum lipids, we also adjusted for change in physical activity. For changes in weight, baseline values of weight, height, and height squared were included. For the model of change in low-density lipoprotein cholesterol, we included change in dietary percentage of energy from saturated fat and change in cholesterol per 4184 kJ. For the model of change in high-density liprprotein cholesterol, changes in percentage of energy from polyunsaturated fat and alcohol were included. The triglyceride model included change in percentage energy from saturated fat and carbohydrate as well as alcohol. A linear contrast statement was used to test for trends across the ordered categories of olestra consumption.
The mean (SD) age of cohort study participants was 41.2 (14.0) years, 62.8% were female, 85.9% were white, 38.1% had a college education, and 28.0% were obese, defined as having a body mass index exceeding 30.13
Table 1 gives associations of olestra intake with dietary change from baseline to the year-1 follow-up. In the cohort as a whole, there was a decrease in total energy, an increase in dietary cholesterol, and only minimal changes in percentage energy from total and saturated fat. Olestra consumption was associated with a statistically significant decrease in percentage energy from total and saturated fat. In the highest olestra intake category, percentage energy from fat decreased 2.7 percentage points, and there was a significant dose-response trend for larger decreases in percentage energy from fat with increasing olestra consumption. There was also a significant trend for decreased percentage energy from saturated fat with olestra consumption. There was no evidence for changes in total energy or cholesterol intake associated with olestra intake.
Table 2 gives associations of olestra intake with changes in body weight and serum lipid concentrations from baseline to the year-1 follow-up. In the entire cohort, there were small decreases in weight and serum total and low-density lipoprotein cholesterol and triglycerides, with an increase in HDL cholesterol. There was a statistically significant decrease in total serum cholesterol associated with olestra intake. In the highest olestra intake category, total serum cholesterol decreased by 0.54 mmol/L (21 mg/dL) (a 10.2% reduction compared with baseline), and there was a significant dose-response trend for larger decreases in total serum cholesterol with increased olestra intake. Although not statistically significant, there was some evidence that olestra intake was associated with weight loss. In the two highest olestra intake categories there was a mean 0.55-kg weight loss, compared with no weight change in nonusers of olestra. Trends in changes in LDL cholesterol paralleled those for total cholesterol but were not statistically significant. There were no associations of changes in HDL cholesterol or triglycerides with olestra consumption.
This report provides the first evidence from a population-based sample of adults that consumption of a fat-modified food (eg, olestra-containing savory snacks) is associated with decreases in fat intake and total serum cholesterol levels. We know of no other studies examining the nutritional effects of new fat-modified foods in the US food supply. This study does indicate that the favorable changes in diet intake generally detected in laboratory-based studies of food consumption are reproduced in "real life" settings.
The association between olestra consumption and serum cholesterol concentrations has several potential, and not mutually exclusive, explanations. Olestra can reduce serum cholesterol concentrations (1) by the inhibition of dietary cholesterol absorption,14 (2) by the replacement of dietary saturated fat, and (3) by the reduction of energy intake. In controlled feeding studies, as little as 3 g/d of olestra could reduce serum concentrations of other lipophilic dietary compounds (eg, serum carotenoids) by over 20%.15 In this report, the heaviest consumers reported 2 g/d or more of olestra, and it is plausible that olestra-containing chips are routinely consumed with significant sources of dietary cholesterol, such as hamburgers, which could reduce absorption of a substantial amount of cholesterol.
Olestra may also be a marker for healthful lifestyle choices. This hypothesis is supported by another report by the OPMSS in which we found that early adopters of olestra-containing savory snacks had strong beliefs in a connection between diet and health.16 For example, compared with individuals who did not consume olestra, consumers were statistically significantly more likely to believe in the importance of eating a low-fat diet and to think about health when selecting foods. These health beliefs influence disease risk by modifying behaviors such as diet and exercise.17 Although we attempted to control for changes in other healthful behaviors (eg, dietary fat intake, physical activity) that might influence serum cholesterol levels, the absence of residual confounding cannot be assured.
Regardless of the mechanisms, the association of olestra consumption and serum cholesterol concentrations warrants further study, particularly since the magnitude of the effect was similar to that seen for dietary fiber18 and even pharmocologic agents.19 For example, in a meta-analysis, Brown et al18 reported that 10 g of soluble fiber per day was associated with a decrease in total cholesterol of −0.45 mmol/L (17 mg/dL), which is comparable to the decrease of −0.40 mmol/L (15 mg/dL) among heavy olestra consumers compared with nonconsumers observed in the present report.
The major limitation of this study is that dietary data, including consumption of olestra-containing savory snacks, are self-reported and therefore likely to contain significant measurement error.20 The FFQ used in this study is the same questionnaire developed for the Women's Health Initiative, a multicenter clinical trial and observational study of approximately 164,500 women.21 In a validity study comparing nutrient estimates from this FFQ with 8 days of food records and recalls,10 correlation coefficients for the nutrients examined in this study were 0.37 for energy, 0.62 for percentage energy from fat, 0.56 for saturated fat, and 0.49 for cholesterol. The reliability of this instrument was good, with intraclass correlation coefficients ranging from 0.69 to 0.73 for these same nutrients. The analyses performed here focus on changes in dietary intake, and therefore systematic bias (eg, underreporting associated with obesity) should not affect our findings. Nonetheless, results based on self-reported dietary intake should be interpreted with caution. In particular, energy intakes from FFQs are not precise.
We note that overall energy intakes estimated from the FFQ decreased by almost 9% from the baseline assessment to year 1, although energy-controlled measures (eg, percentage of energy from fat) were generally stable. Neither the FFQ nor the data collection procedures were significantly changed from baseline to year 1; therefore, we can only speculate that there may have been a cohort effect resulting in poorer completion of the questionnaire (eg, a fatigue effect). We also note that the sample size for olestra consumers, particularly those in the highest category of consumption, is small. Finally, although our response rates were good for a clinic-based study, these results may not be generalized to other populations.
Our results indicate that olestra is associated with healthful changes in dietary fat intake and serum cholesterol concentrations in a population-based sample of American consumers. This finding counters suggestions in the literature that consumption of fat-reduced foods results in increased consumption of other, possibly less desirable foods. There has been considerable concern that olestra consumption will result in reduced concentrations of serum fat-soluble vitamins and carotenoids and may cause gastrointestinal cramping and/or diarrhea. In a separate investigation of this same population, we observed no statistically significant trends in reductions in serum concentrations of vitamins A, D, E, or K or carotenoids associated with olestra consumption.22 While controversy continues regarding whether olestra causes serious gastrointestinal effects, a recent double-blind feeding among 563 free-living adults found no evidence that consumption of olestra-containing potato chips was associated with more gastrointestinal symptoms of any type than regular, triglyceride-containing chips.23 Additional follow-up of the full-scale OPMSS will provide larger samples and therefore additional power for more detailed investigations of changes in eating patterns and other health indicators, particularly in different sex, race/ethnicity, and socioeconomic groups.
Accepted for publication April 6, 2000.
This study was funded by Procter & Gamble, Cincinnati, Ohio.
Corresponding author: Ruth E. Patterson, PhD, RD, Cancer Prevention Research Program, Fred Hutchinson Cancer Research Center, 1100 Fairview Ave N, MP-702, Seattle, WA 98109 (e-mail: firstname.lastname@example.org).