Are awareness of dietary fat intake and actual fat ... - Nature

European Journal of Clinical Nutrition (1997) 51, 542±547 ß 1997 Stockton Press. All rights reserved 0954±3007/97 $12.00

Are awareness of dietary fat intake and actual fat consumption associated?ÐA Dutch±American Comparison K Glanz1, J Brug2, and P van Assema2 1

Cancer Research Center of Hawai'i, University of Hawai'i, Honolulu, HI, USA; and 2 Department of Health Education and Health Promotion, Maastricht University, Maastricht, the Netherlands

Objectives: To compare dietary fat intake, the accuracy of individuals' awareness about their fat intake, and sociodemographic and psychosocial correlates of awareness, in Dutch and American samples of employed adults. A discrepancy between objective dietary intake data and subjective self-evaluation of dietary fat consumption has been recognized in the past and might undermine healthy diet promotion interventions, and this is important because people who believe that their diets are healthful are less likely to be interested in making changes. Further, international comparisons have not been examined to date. Design: Data collected for the baseline surveys of the `Healthy Bergeijk' study in the Netherlands and the `Working Well Trial' in the United States were compared. Subjects: Working adults from a Dutch community health intervention study (n 768) and an American worksite health promotion trial (n 15 440). Main outcome measures: Objectively assessed dietary fat intake, measured by food frequency questionnaires, and subjective ratings of fat intake (self-rated fat intake). Results: Findings show that the Dutch respondents had higher objectively assessed fat intake and lower subjective ratings of fat intake (P < 0.001). American respondents perceived their diets as higher in fat, more often stated their intentions to reduce fat intake, and were slightly more likely to make realistic estimates of their dietary fat. Dutch subjects were signi®cantly more likely to underestimate their fat intakes. In both samples, women were most likely to underestimate their fat consumption and the most educated persons were most likely to be realistic. Conclusions: A substantial proportion of adults, both in the United States and the Netherlands, lack accurate awareness about how much fat they consume, though errors tend to be in opposite directions in the two countries. This study is an important ®rst step toward broadening our international understanding of human dietary behavior for disease prevention. Sponsorship: The Working Well Trial was supported by a Cooperative Agreement from the National Cancer Institute, grants no. U01 CA51671, U01 CA51686, U01 CA51687, U01 CA 51688, and P01 CA 50087. The Healthy Bergeijk project was supported by a grant from the Dutch Cancer Society. Descriptors: dietary fat; nutrition education; fat consumption; dietary behavior change; international health promotion

Introduction The growing epidemiological evidence of a relationship between dietary patterns and various chronic diseases, including cardiovascular (CVD) disease and some cancers, has led governments and public health organizations to promote preventive dietary recommendations (National Research Council, 1989; Surgeon General's Report, 1988). In the United States the most recent guidelines for a healthy diet were published in 1995 (USDA/USDHHS, 1995). These are similar to guidelines that are used in European countries and to recommendations for desirable dietary changes to prevent chronic diseases such as CVD and cancer (Benito, 1992; Butrum et al, 1988; American Heart Association, 1988; NRC, 1989). In the Netherlands recommendations for a prudent diet were published by the Dutch Nutrition Council in 1986 (Dutch Nutrition Council, Correspondence: Dr K Glanz, Cancer Research Center of Hawai'i, University of Hawai'i, 1236 Lauhala Street, Suite 406, Honolulu, Hawai'i 96813, USA. Received 27 November 1996; revised 28 March 1997; accepted 17 April 1997

1986). These recommendations are consistent with the most recent dietary recommendations in the United States, but differ in some details. The US dietary guidelines advise reducing dietary fat intake to 30% or less of total calories while in the Netherlands 35% is the upper limit for desirable fat intake. Another difference is that US guidelines now include speci®c quantitative advice for fruit and vegetable intake (Havas et al, 1994). Major discrepancies between the recommendations and results of food consumption research have been found in both the US and the Netherlands (Hulshof, 1993; Block, 1991). As a result, public health initiatives have been organized to promote population-wide adoption of more health-enhancing dietary patterns in both countries (Riedstra et al, 1993; Brug et al, 1993; Van Assema et al, 1993; Butrum, 1988; NCEP, 1990). Nutrition interventions aimed at reducing fat consumption have gained especially broad attention. In the Netherlands, the government-sponsored Nutrition Council recommended a focus on dietary fat reduction as the major objective in healthy diet promotion interventions, because the largest health gain was expected from this type of change (Dutch Nutrition Council, 1986).

Awareness of dietary fat intake K Glanz et al

The fat reduction message also has received broad attention in the US (Butrum et al, 1988; NCEP, 1990), though recently there has also begun to be a strong focus on promoting fruit and vegetable consumption (Havas et al, 1994). At present, fat consumption in the Netherlands is somewhat higher than fat consumption in the US (Hulshof et al, 1993). Further, research in the Netherlands has shown that the majority of Dutch adults consume a diet that is too high in fat according to the Dutch recommendations. Nevertheless, only a small minority of the Dutch population, approximately 10%, rate their diets as too high in fat. This discrepancy between objective dietary intake data and subjective self-evaluation of dietary fat consumption might be a major barrier in healthy diet promotion interventions aimed at reducing dietary fat intake. People who believe that they are not consuming too much fat in their diets are less likely to be interested in reducing dietary fat intake (Brug et al, 1993; Brug et al, 1994). Furthermore, relatively low correlations between objectively assessed dietary fat intake and subjects' self-rated dietary fat intake have been found in the Netherlands as well as the US (Glanz et al, 1993; Brug et al, 1994). There are reasons to believe that lack of awareness of dietary fat intake might be less of a concern in the US than in the Netherlands. In the US there has been an interest among the general public in healthy nutrition, and especially reduction of fat intake, for a longer period of time (Samuels, 1990; Samuels, 1993). Also, nutrition education aimed at the general public was initiated earlier in the US than in the Netherlands. Consumer surveys show that dietary fat has recently become the greatest nutritional concern among Americans (FMI, 1992). Beyond these observations made in separate research reports from the US and the Netherlands, it is possible to learn more by comparing awareness of dietary fat intake and its psychosocial correlates internationally. In this article Dutch and American data on objectively assessed dietary fat intake and on self-rated dietary fat intake are compared, in order to (1) study differences in awareness of dietary fat consumption between the Netherlands and the United States, and to (2) evaluate whether awareness is associated with intentions to reduce dietary fat intake, social support for healthy eating, and subjects' background characteristics. It was hypothesized that a higher proportion of subjects in the Netherlands than in the United States would be unaware of their levels of dietary fat intake. It was also hypothesized that high self-rated dietary fat intake would be more common in the US, and thus the proportion of subjects intending to reduce dietary fat intake would also be higher in the US. Methods In this study data collected for the baseline survey of the `Healthy Bergeijk' study conducted in the Netherlands were compared to data from the baseline survey of the `Working Well Trial' in the United States. Ethical approval was obtained from the Institutional Review Boards at all participating institutions before data collection was undertaken. Healthy Bergeijk The Healthy Bergeijk Project was a community based cancer prevention intervention project funded by the Dutch Cancer Society. The aims of the project were to

develop and implement an intervention to reduce the prevalence of four cancer-related risk behaviors (smoking, high fat consumption, high alcohol intake, use of solariums) among the population of Bergeijk, a municipality in the south of the Netherlands. The intervention was evaluated by means of a pretest±posttest control group design with two posttests. Details regarding the intervention and the evaluation study are reported elsewhere (Van Assema et al, 1993). Data for evaluation of the Healthy Bergeijk project were collected by telephone interviews. To evaluate dietary behavior change, data were collected on: objectively assessed dietary fat intake, self-rated dietary fat intake, intentions to reduce dietary fat intake, and psychosocial correlates of dietary fat reduction. Objectively assessed dietary intake was measured by means of a validated 25 item food frequency questionnaire that yields a fat consumption score between 12 and 60. The score is based on total fat intake and allows rank-ordering of respondents with regard to their fat consumption (Van Assema et al, 1992). Self-rated intake was measured by asking respondents to rate how the fat content of their diets on a bipolar ®ve-point scale (from very low in fat to very high in fat). Psychosocial correlates of dietary fat intake were measured by a questionnaire that was speci®cally developed for the Healthy Bergeijk study using a combination of qualitative and quantitative techniques (Van Assema et al, 1992). The constructs in the questionnaire were based on a combination of the Theory of Planned Behavior and Social Cognitive Theory (Ajzen & Madden, 1986; Bandura, 1986), and include attitudes, intention to change, perceived social in¯uence, self-ef®cacy, and social norms. Speci®c information on the development of this questionnaire has been reported elsewhere (Van Assema et al, 1993; DeVries et al, 1992), and an English translation of the item wording appears in Appendix A. For the present study data from the baseline survey of the Healthy Bergeijk study were used. The survey was conducted in February of 1990. The study sample consisted of 1507 subjects, randomly selected from the municipality population records in Bergeijk and a randomly selected comparison community. The 751 men and 756 women with completed interviews represented a response rate of 75.5%. In this analysis, only respondents who reported that they were employed at the time of the interview (n 768) were included in the analysis, to improve the comparability to the Working Well survey respondents. Working Well The Working Well Trial, a ®ve year cooperative agreement funded by the National Cancer Institute, tested the effectiveness of worksite health promotion interventions in achieving individual and organizational changes to reducing cancer risk. Working Well is a randomized, prospective ®eld experiment with 114 worksites and 37 291 workers. The study includes companies and workers with broad geographic and industrial diversity in the United States, and was conducted by four project study centers, a data coordinating center, and the National Cancer Institute. All Working Well study centres targeted nutrition and at least one other prevention component, for example, smoking, cancer screening, occupational health. The study is described in detail elsewhere (Abrams et al, 1994). The evaluation of Working Well was based on surveys of individual workers using self-administered questionnaires. The survey instrument was a self-administered

543


544

questionnaire with questions about respondents' sociodemographic characteristics; knowledge, attitudes, and behaviors related to nutrition and smoking; and perception of the worksite. Dietary intake was measured with an 88-item semi-quantitative food frequency questionnaire (FFQ) with portion sizes (176 items total) (Kristal et al, 1994). The FFQ analysis software is based on a nutrient database developed by the University of Minnesota Nutrition Coordinating Center (Schakel et al, 1988). Both total fat intake and percent energy from fat were calculated. For this analysis the main dietary variable of interest was the total fat intake, in grams. Fat intake was categorized into tertiles by gender, for comparability to the fat intake measure from the Healthy Bergeijk study. The survey also included measures of key psychosocial factors believed to in¯uence food choices in three domains: predisposing, enabling, and change-related factors. These factors were based on several theoretical frameworks that were foundations for the intervention, including Social Cognitive Theory (Bandura, 1986); diffusion of innovations (Rogers, 1983), community organization (Minkler, 1990), Theory of Planned Behavior (Ajzen & Madden, 1986), and the stages of change model (Prochaska et al, 1992). The procedures used for developing these measures and establishing their criterion validity are described in detail elsewhere (Glanz et al, 1993; Glanz et al, 1994). Four psychosocial factors were of particular interest in the present analysis: self-rated fat intake, intentions to reduce dietary fat, perceived support for a healthy diet from family/friends, and perceived support for healthy eating from colleagues (see Appendix A). The Working Well baseline survey was conducted in the fall of 1990, with survey administration by mail or in group meetings at worksites. A total of 20 801 respondents completed and returned the baseline survey for a 69% overall response rate; response rates for each of the 114 worksites ranged from 23±98% (Heimendinger et al, 1995). To reduce the potential for non-response bias in the present analysis, we include only responses from the 91 worksites that obtained response rates of 55% or greater (mean 78%), for a total of 17 118 employees. Statistical analysis In both the Bergeijk study and Working Well, objectively assessed dietary fat intake (òbjective' dietary fat intake), self-rated dietary fat intake (subjective dietary fat intake), intention to reduce dietary fat intake (intention), and encouragement from the social environment to reduce dietary fat intake (social support) were measured. Responses of subjects in both study populations were divided in tertiles by gender for objectively assessed dietary fat intake. The tertiles were de®ned as low, intermediate, or high fat diets. Subjects were also divided into three groups de®ned as low or very low, intermediate, and high or very high in fat, for self-rated dietary fat intake. Parallel analyses were conducted for the Dutch and American samples. For analyses of the Working Well (American) sample, we adjusted for the worksite cluster sampling effect using Rao and Scott's method for chi-square tests (Rao & Scott, 1985). In order to make a comparison in dietary fat intake awareness between Dutch and American subjects an awareness variable was created by comparing objectively assessed dietary fat intake with subjective dietary fat intake. Subjects were classi®ed as good estimators if their subjective dietary fat intake was in the same category as their objectively assessed dietary fat intake. Subjects were

classi®ed as underestimators if their subjective intake was lower than their òbjective' fat intake. Subjects were classi®ed as overestimators if their subjective dietary fat intake was higher than their òbjective' dietary fat intake. For both study populations the association of self-rated dietary fat intake with objectively assessed fat intake and with intention to reduce dietary fat intake were assessed. Chi-square tests were used to study differences in proportions of subjects who are good estimators, underestimators, and overestimators of their fat intake, between the Healthy Bergeijk and Working Well study populations. Next, we examined the association of demographic characteristics (age, gender, and education) and psychosocial factors (intentions to change, and social support) with accuracy of dietary fat self-ratings, or association between self-rated and objectively assessed fat intake. Results Table 1 describes the background characteristics, fat intake, and psychosocial variables for the two study samples. Both samples were approximately two-thirds male, with about 80% of subjects below age 50, and with more than 40% of subjects having education beyond high school. Respondents in Bergeijk were somewhat younger and less educated. Dietary fat intake, though not measured as percent energy from fat in the Dutch sample, was estimated at 40.5%en in Healthy Bergeijk (based on results of a comprehensive nationwide food consumption survey conducted during the same period) and 36.7%en in Working Well (data not shown). Signi®cantly more Dutch subjects rated their fat intake as low, and more Working Well subjects rated their fat intake as moderate. Working Well respondents were signi®cantly more likely to intend to adopt a lower-fat diet within the next six months, and social support for low-fat eating from friends/family and from co-workers was signi®cantly higher among Working Well subjects. Table 2 gives the proportion of respondents in each category of self-rated fat intake vs objectively assessed fat intake, and Table 3 presents the proportions of subjects in the different awareness (èstimators') categories for both study populations. In the Dutch study 42.9% of all subjects made realistic estimates of their dietary fat intake, and in the American study 45% of subjects were realistic, or good estimators. 36.5% of Dutch respondents underestimated their fat intake and 27.4% of American respondents were underestimators. In contrast, 27.6% of the US respondents overestimated their fat intake compared to only 20.6% of Dutch respondents. Differences between the two samples were highly statistically signi®cant (P < 0.001). Table 4 presents the associations between demographic characteristics and the accuracy of self-rating dietary fat intake (awareness), for the Healthy Bergeijk and Working Well samples. There was no signi®cant association between awareness and age in either sample. Males were more likely to overestimate and less likely to underestimate in both samples (P < 0.001). Highly educated persons were most often realistic, or good estimators, in both samples. However, the least educated Dutch group and the most educated American group were found to be underestimators (P < 0.001). Table 5 shows the associations between psychosocial factors and awareness for the Dutch and US samples. Intentions to adopt a low-fat diet in the next six months was much higher overall among Working Well respondents, and patterns of intention in relation to awareness


545

Table 1 Description of study samples: demographics, fat intake, and psychosocial variables N

Working Well (15440)a

Healthy Bergeijk (768)

Demographic characteristics Gender: % male Age < 35y 35±50y 50y Education 12y 13±15y 16y Dietary fat intake Fat intake Self-rated fat intake (%) Low Middle High

67.8

68.1

43.1% 41.9% 15.0%

32.3% 45.5% 22.3%

58.5% 24.7% 16.5%

49.3% 32.8% 17.8%

28.7b

72.1 (39.2)c

46.0% 32.4% 21.6%

22.3% 55.6% 22.0%

6.9% 12.2% 80.9%

26.4% 54.0% 19.6%

58.6% 25.9% 15.5%

20.3% 53.7% 26.1%

88.3% 6.0% 5.7%

59.1% 36.2% 4.6%

Psychosocial variables: intentions and social support Intention to adopt low-fat diet within 6 months (%) Yes Maybe No Social support from friends and family None Some A lot Social support from co-workers None Some A lot a b c

Removed respondents who did not self-rate diet from analyses (n 1778). Fat intake measured as a score ranging from 12±60, not in grams. Fat intake in grams (mean s.d.).

Table 2 Self-rated fat intake vs objectively assessed fat intakea Objectively assessed fat intake Healthy Bergeijk

Self-rated fat intake Low Total % Row % Medium Total % Row % High Total % Row %

Working Well

Low

Medium

High

Low

Medium

High

21.1 46.2

15.2 33.1

9.5 20.7

13.0 58.2

6.4 28.5

3.0 13.3

10.0 30.9

10.9 33.7

11.9 35.3

17.5 31.5

20.2 36.2

18.0 32.3

4.2 19.3

6.5 30.1

10.9 50.6

3.5 15.8

6.7 30.6

11.8 53.6

Boldface ®gures are good estimators. Chi-square signi®cant (P < 0.001).

a

Table 3

Accuracy of self-ratings of fat intakea Accuracy of ratings

Underestimators Good estimators Overestimators a

Healthy Bergeijk

Working Well

36.5% 42.9% 20.6%

27.4% 45.0% 27.6%

Chi-square signi®cant (P < 0.001).

were in the opposite direction in the two samples. Among the Dutch, a trend was found in which people who overestimated their actual fat intake were most likely to express intentions to change; but among the Americans, intentions were highest among underestimators. With respect to social support, the only trend found was in the Healthy Bergeijk

sample, in that good estimators and underestimators perceived greater support from family and friends for consuming lower-fat diets (P 0.08). Social support from coworkers was not associated with awareness of fat intake in either sample. Discussion In this study we compared dietary fat intake, the accuracy of individuals' awareness about their fat intake, and sociodemographic and psychosocial correlates of awareness, in Dutch and American samples of employed adults. As we hypothesized, the Dutch respondents had higher objectively assessed fat intake and lower subjective ratings of fat intake. The American sample perceived their diet as higher in fat and more often stated their intentions to


546

Table 4 Demographics and association between self-rated and objectively assessed fat intake Gender (% male)a

Age (mean)

Underestimators Good estimators Overestimators a

Education (% high)a

Healthy Bergeijk

Working Well

Healthy Bergeijk

Working Well

Healthy Bergeijk

Working Well

36.5 37.6 35.9

39.5 41.0 41.2

60.4 65.6 85.5

64.7 68.3 71.9

12.6 19.6 17.0

19.6 18.9 16.1

Chi-square signi®cant for both Healthy Bergeijk and Working Well samples (P < 0.001).

Table 5 Psychosocial factors and association between self-rated and objectively assessed fat intake Intentiona (% de®nitely/maybe)

Underestimators Good estimators Overestimators a b

Social support: family and friendsa (% a lot/some)

Social support: co-workers (% a lot/some)

Healthy Bergeijk

Working Well

Healthy Bergeijk

Working Well

Healthy Bergeijk

Working Well

17.2 18.3 24.3

48.6 42.2 39.8

46.4 40.5 34.6

80.2 80.0 79.2

12.2 12.1 10.1

40.6 41.4 40.4

Chi-square borderline signi®cant (P 0.08) for Healthy Bergeijk. Chi-square borderline signi®cant (P < 0.001) for Working Well.

reduce fat intake. US respondents were slightly more likely to make realistic estimates of how high in fat their diets were, but Dutch subjects were signi®cantly more likely to underestimate their fat intakes. In both samples, women were most likely to underestimate their fat consumption and the most educated persons were mostly likely to be good estimators. This latter association was strongest in the Bergeijk subjects; in the Working Well sample the most educated were more likely to be realistic than to overestimate fat intake, and still more likely to underestimate. Paradoxically, the patterns of awareness and intentions to change were in the opposite directions in the Dutch and American samples: Dutch overestimators and American underestimators most often said they planned to try to reduce dietary fat. However, this ®nding should be viewed in context: American respondents were about four times as likely to plan to make changes as were the Dutch subjects. Thus adults in the Netherlands who believe their diets are high in fat plan to initiate changes, while American adults who already believe they are following low-fat eating plans (perhaps erroneously) say they intend to begin, or perhaps continue, trying to change. It is clear that a substantial proportion of adults, both in the US and the Netherlands, lack accurate awareness about how much fat they consume, though errors tend to be in opposite directions in the two countries. A possible reason for the lack of awareness is the lower prevalence of obesity in the Netherlands, where overweight status may be confused with èating too much fat'. The relative recency of public health nutrition campaigns targeting fat reduction are another possible explanation. In contrast, obesity is increasing in the US and Americans express great concern with dietary fat (FMI, 1992). This, along with aggressive marketing of reduced fat food products in the US, may lead Americans to believe that they should continually try to lower their fat intake irrespective of the current level. Importantly, in both countries (and likely in other industrialized nations as well), we need to recognize that knowledge about the nutritional value of one's diet is often ¯awed, and that perceptions more than actual eating habits are most likely to lead to motivation to change (Brug et al, 1994; Glanz et al, 1994). Thus, giving accurate feedback may stimulate interest in undertaking dietary improvements. However, some people may be distressed

to learn that their diets are unhealthy (Bowen et al, 1994), and health educators need to remain alert to this possibility. This research was possible because of the concurrent undertaking of research addressing similar constructs in comparable adult, employed samples in the Netherlands and the US. This also contributed to limitations of our analyses: the studies differed in their data collection methods (telephone and mail), measures of fat intake (rank ordering vs quanti®able), language of questions (Dutch and English) and exact wording of speci®c items, and samples sizes. It further precluded more in-depth analyses on a broader set of variables. In the future, planned cross-national studies should be designed a priori with the aim of helping scientists understand how social and public health conditions affect health promotion efforts. This study is an important ®rst step toward broadening our international understanding of human dietary behavior for disease prevention. AcknowledgementsÐThe authors wish to acknowledge the contributions of Ruth Patterson, Alan Kristal, and Steve Thomson, who assisted in analysis of the Working Well data. Appendix A Item wording: Working Well trial and Healthy Bergeijka Variable

Item wording Working Well

Healthy Bergeijk

Self-rated intake

How high in fat is your overall diet? . . . very high, high, moderate, low, very low

Intention

Do you plan to cut down on the fat in your diet? How much encouragement for eating low-fat foods do you get from your coworkers? How much encouragement for eating low-fat foods do you get from your family and close friends?

Do you consider your diet quite high in fat, high in fat, in the middle, low in fat, or quite low in fat? Do you plan to start eating less fat . . . ?

Social support from co-workers

Social support from family and friends

a

Translated into English.

Do your colleagues encourage you to eat a diet with less fat? Do your partner and close relatives encourage you to eat a diet with less fat?


References Abrams DB, Boutwell WB, Grizzle J, Heimendinger J, Sorensen G & Varnes J (1994): Cancer control at the workplace: The Working Well Trial. Prev. Med. 23, 1±13. Ajzen I & Madden TJ (1986): Prediction of goal-directed behavior. Attitudes, intentions and perceived behavioral control. J. Exper. Soc. Psychol. 22, 435±474. American Heart Association (1988): Dietary guidelines for healthy American adults. Circulation 77, 721A±724A. Bandura A (1986): Social Foundations of Thought and Action. A Social Cognitive Theory. Englewood Cliffs (NJ): Prentice-Hall. Benito E (1986): Overview of dietary recommendations on diet and cancer. In Public Education on Diet and Cancer, eds. E Benito, A Giacosa, MJ Hill, Kluwer: Dordrecht. Block G (1991): Dietary guidelines and the results of food consumption surveys. Am. J. Clin. Nutr. 53, 356S±357S. Bowen DJ, Fries E & Hopp HP (1994): Effects of dietary feedback on behavioral and psychological variables. J. Behav. Med. 17, 589±604. Brug J, Riedstra M, Pruyn J & LoÈwik MRH (1993): De effecten van de derde landelijke Let op Vet campagne (Effects of the third Fat Watch campaign) (in Dutch). Voeding 54, 8±10. Brug J, Van Assema P & Kok GJ (1995): Psychosocial determinants of dietary fat intake and intention to change. ErnaÈrungs-Umschau 42 (suppl), 150±152. Brug J, Van Assema P, Lenderink T, Glanz K & Kok G (1994): Self-rated dietary fat intake: association with objective assessment of fat, psychosocial factors, and intention to change. J. Nutr. Educ. 26, 218±223. Butrum RR, Clifford CK & Lanza E (1988): NCI dietary guidelines: rationale. Am. J. Clin. Nutr. 48, 888±895. De Vries H, Dijkstra M & Kuhlman P (1988): Self-ef®cacy: The third factor besides attitudes and subjective norm as a predictor of behavioral intentions. Health Educ. Res. 3, 273±282. De Vries H, Weijts W, Dijkstra M & Kok G (1992): The utilization of qualitative and quantitative methods for health education program planning, implementation and evaluation: a spiral approach. Health Educ. Q. 19, 101±115. Dutch Nutrition Council (Voedingsraad) (1986): Advice on Dietary Recommendations. Voedingsraad: Den Haag. Edts Can®eld, Krinsky and Olson (1993): Published by the New York Academy of Sciences ANYAA, 691, 1±300. Farugue MD, Khan MR, Rahman MD & Faruk A. (1995): Relationship between smoking and antioxidant nutrient status. B. J. Nutr., 73, 625± 632. Food Marketing Institute (1992): Trends: Consumer Attitudes and the Supermarket. Washington, DC: Author. Glanz K, Kristal AR, Sorensen G, Palombo R, Heimendinger J & Probart C (1993): Development and validation of measures of psychosocial factors in¯uencing fat- and ®ber-related dietary behavior. Prev. Med. 22, 373±387. Glanz K, Patterson RE, Kristal AR, DiClemente CC, Heimendinger J, Linnan L & McLerran DF (1994): Stages of change in adopting healthy diets: Fat, ®ber, and correlates of nutrient intake. Health Educ. Q. 21, 499±519. Havas S, Heimendinger J, Reynolds K, Baranowski T, Nicklas T, Bishop D, Buller D, Sorensen G, Beresford S, Cowan R, Damron D (1994): A Day for Better Health: A new research initiative. J. Amer. Diet. Assoc. 94, 32±36. Heimendinger J, Feng Z, Emmons KM, Stoddard A, Kinne S, Biener L, Sorensen G, Abrams DB, Varnes J & Boutwell B for the Working Well Research Group (1995): The Working Well Trial: Baseline dietary and smoking behaviors of employees and related working characteristics. Prev. Med. 24, 180±193. Hertog MG, Hollnon PC. & Putte Van de B. (1993) Content of potentially anticarcinogenic ¯avonoids of tea, infusions, wines and fruit juices. J. Agric. Food. Chem. 41, 1242±1245.

Hulshof KAFM, LoÈwik MRH, Kistemaker C, Hermus RJJ, Ten Hoor F & Ockhuizen Th (1993): Comparison of dietary intake data with guidelines: some potential pitfalls (Dutch Nutrition Surveillance System). J. Amer. Coll. Nutr. 12, 176±185. Kristal AR, Shattuck AL & Williams AE (1994): Food frequency questionnaires for diet intervention research. In: Proceedings of the 17th National Nutrient Databank Conference, June 1992. Washington DC, International Life Sciences Institute, 110±125. Minkler M (1990): Improving health through community organization. In Health Behavior and Health Education: Theory, Research and Practice, eds. Glanz K, Lewis FM, Rimer BK. San Francisco: Jossey-Bass. National Cholesterol Education Program (1990): Report of the expert panel on population strategies for blood cholesterol reduction. Bethesda, MD: National Institutes of Health. (NIH Publ. No. 90-3046). National Research Council, Committee on Diet and Health (1989): Diet and Health: Implications for Reducing Chronic Disease Risk. Washington, DC: National Academy Press. Omenn GS, Goodman GE, Thornquist MD, Balmes J, Gullen MR, Glass A, Keogh JP, Meyshens FL, Valanis B, Williams JH, Barnhart S, Cherniack MG, Brodhin CA, Hamman S. (1996): Risk factors for lung cancer and for intervention effects in CARET, the beta-carotene and retinol ef®cacy-trial. J. Natl. Cancer Inst., 88, 1550±9. Prochaska JO, DiClemente CC & Norcross JC (1992): In search of how people change: Applications to addictive behaviors. Amer. Psychol. 47, 1102±1114. Rao JN & Scott JA (1985): On chi-squared tests for multiway contingency tables with cell properties estimated from survey data. Annals. Statist. 12, 46±60. Riedstra M, Brug J, Hardeman W, Pruyn JFA, LoÈwik MRH (1993): Three years national Fat Watch campaign (1991±1993): What are the effects (in Dutch). Voeding 54, 4±7. Riedstra M, Hardeman W, Pruyn JFA, Van der Feen de Lille JCJF & Jonkers R (1992): The Effects of the Second National Fat Watch Campaign in 1992 (in Dutch). TNO-report V 92.453. TNO Nutrition Institution, Zeist, The Netherlands. Rogers EM (1983): Diffusion of Innovations (3rd Ed.). New York: The Free Press. Samuels SE (1990): Project LEAN: A national campaign to reduce dietary fat consumption. Am. J. Health Promot. 4, 435±440. Samuels SE (1993): Project LEANÐLessons learned from a national social marketing campaign. Publ. Health Rep. 108, 45±53. Schakel SF, Sievert YA, Buzzard IM (1988): Sources of data for developing and maintaining a nutrient database. J. Amer. Diet. Assoc. 88, 1260±1271. Stahl W (1996): Desirable versus potentially harmful intake levels of various forms of carotenoids. In ``Natural antioxidants and food quality in atherosclerosis and cancer prevention. J. Kumpulainen and JT Salonen, Edts, Royal Society of Chemistry. US Department of Agriculture, US Department of Health and Human Services (1995): Nutrition and Your Health: Dietary Guidelines for Americans (4th Ed.) Washington, DC: US Governmental Printing Of®ce. US Department of Health and Human Services, Public Health Service (1988): The Surgeon General's Report on Nutrition and Health. Washington, DC: Government Printing Of®ce. DHHS publication no. (PHS) 88-50210. Van Assema P, Brug J, Kok G & Brants H (1992): The reliability and validity of a Dutch questionnaire on fat consumption as a means to rank subjects according to individual fat intake. Eur. J. Cancer Prev. 1, 375± 380. Van Assema P, De Vries H & Kok G (1992): Cancer prevention research by means of focus group interviews (in Dutch). Gedrag en Gezondheid. 20, 10±22. Van Assema P, Pieterse M, Kok GJ, Eriksen M &De Vries H (1993): The determinants of four cancer related risk behaviors. Health Educ. Res. 8, 461±472.

547