5 A Day for Better Health Program Evaluation Report: Evaluation

Implementation

This section assesses the implementation of the 5 A Day Program and addresses two questions: (1) to what degree were resources made available to the Program and (2) to what extent were the components of the Program implemented as conceptualized and proposed to NCI's Board of Scientific Counselors?

Media Campaign

As a joint effort between the NCI/OCC and the PBH, the media component of the 5 A Day Program was implemented as proposed. Using the Consumer-Based Health Communications Model, formative research was conducted, the 5 A Day target audience selected, and communications strategies and messages designed and implemented. By 1997, $5.15 million ($5 million proposed) had been directed to a national media campaign that spread the 5 A Day message through media events and activities to improve public awareness. Since 1997, the NCI has allocated another $2.6 million to continue the national media campaign.

Community-Level Interventions

The community component of the 5 A Day Program underwent a shift in emphasis as the program was implemented. Originally planned as a channel to create state and local-level interventions in collaboration with local industry partners and community groups, the emphasis shifted from community/state capacity building to a more research-oriented agenda. The NCI allocated the original $16 million proposed for state/community support through the RFA process to university-based investigator research to develop and evaluate community-based interventions. This shift strengthened the opportunities to test well-designed intervention strategies for specific channels and targeted populations, but left little support for capacity building at the state and community level. Although all states eventually became licensed by the 5 A Day Program and used educational materials developed by the NCI and the PBH for national distribution, few states had adequate resources to support or evaluate the program. In 1998, about two-thirds of the states reported devoting less than one full-time equivalent (FTE) to 5 A Day Program activities. When grant funds and expenditures on personnel are ignored, the estimated expenditures for the 5 A Day Program were $50,000 or less per year for most states. However, 31 grants to states were allocated through a partnership between the NCI and the CDC to support and evaluate state health agency programs. These grants totaled $1.75 million from 1992 to 1997 and an additional $1.15 million since 1997.

Point-of-Purchase Program and Industry Partnerships

Agreements between the NCI and the PBH (Memoranda of Understanding, License Agreements) about national structure, organizational roles, how the program would operate, and criteria/guidelines for using the 5 A Day logo/service mark were developed and signed. The NCI and the PBH established a positive working relationship with producers and retailers and implemented the industry component of the 5 A Day Program as planned. By 1994, more than 1,000 licensed partners, representing 35,000 point-of-purchase locations, actively participated in the program. These partners made in-kind contributions totaling an estimated $368 million from 1992 to 1999. Promotions and educational programs engaging consumers in practical methods to increase vegetable and fruit consumption were produced for food service organizations and retail promotions. The beneficial outcomes of the partnership include an expanded communication base for the 5 A Day message and the promotion of national nutritional objectives. This public-private partnership, with its identifiable structure and modules, represents a model for the implementation of other public health endeavors.

Other Partnerships

The 5 A Day Program developed successful collaborations with a range of Federal, state, and voluntary agencies. These collaborations provided mechanisms whereby the 5 A Day message was incorporated into a range of programs, from the school lunch program to statewide public-health interventions. As a result of these efforts, the 5 A Day message is highlighted in the USDA modifications in the school lunch program and the Team Nutrition campaign, the dietary guidelines of the ACS, the Department of Defense health promotion programs, the Indian Health Service nutrition and dietetics programs, and the USDA food assistance programs (Food Stamps; Special Supplemental Nutrition Program for Women, Infants, and Children (WIC); and the Child Care Food Program). In addition, the food industry has modified products and materials to fit the 5 A Day message criteria. The message is generally accepted in the United States and incorporated into most nutrition programs. This is the kind of modification of norms that most programs hope to create.

Research

A research component was added to the original implementation plan for the 5 A Day Program as the original RFA process developed. The NCI funded nine behavior-change research and evaluation studies to determine the effectiveness of 5 A Day interventions. These projects were conducted in community settings (e.g., schools, churches, worksites) and accounted for $18 million in funding in 1992-1997. The NCI expended another $9.6 million after 1997 on investigator- initiated behavioral change research focusing on 5 A Day. The results of these studies have been published in peer-reviewed journals as multiple-site collaborative papers and as reports from individual study sites.

Central Capacity

A 5 A Day program director was hired, and various staff from NCI's OCC performed the functions of the nutrition specialist. The position of evaluation manager was never filled, and the $6 million for contractual administrative support was never allocated to the program by the NCI.

Table 2 lists the actual expenditures of the 5 A Day Program from its inception in 1992 through 1999. For the original 5-year plan, 1992-1997, the actual expenditure of $26.6 million was very close to the goal of $27 million. Total expenditures for the 5 A Day Program were $40.41 million, with the majority of funds ($27.6 million) having been allocated for behavior-change research.

Table 2. Actual NCI cost by fiscal year (dollars in millions)
	Nutrition and Behavior-Change Research (RFA)	State Health Agency Research	Media (Including OCC)	Program Evaluation	Total
FY 1992	—	—	$0.40	—	$0.40
FY 1993	$4.00a	—	$1.00	—	$5.00
FY 1994	$4.00 a	$0.30	$1.00	—	$5.30
FY 1995	$4.00 a	$0.40	$1.00	$0.68	$6.08
FY 1996	$4.00 a	$0.50	$1.00	$0.66	$6.16
FY 1997	$2.00 a	$0.55	$0.75	$0.42	$3.72
FY 1998	$4.00b	$0.50	$1.50	$0.25	$6.25
FY 1999	$5.60 b	$0.65	$1.10	$0.15	$7.50
Total	$27.60	$2.90	$7.75	$2.16	$40.41

^a Supported by funds for RFA.
^b Supported by funds for investigator-initiated research.

Process Measures

This section assesses the Program's impact through its various media components. It answers the overall question: "To what extent have the mass media communicated the 5 A Day message since the Program began?"

Specifically, this section answers the following questions:

• When did the 5 A Day message first appear in the media?
  
  
• What are the national trends in media communication of the message?
  
  
• What are the trends in local media communication, especially in areas covered by the state/community coalitions?
  
  
• To what extent has the public responded to the basic message?

When Did the 5 A Day Message First Appear in the Media?

The media rollout of the 5 A Day Program occurred officially in September 1991. However, the concept of eating at least 5 servings of vegetables and fruit each day has been an established dietary guideline since it was first recommended by the USDA in 1916.⁸¹ The general dietary recommendation to consume larger amounts of vegetables and fruit has varied little through the years, with the exception of the specific number of servings, which has varied from 3 to 5. Beginning in 1980, the USDA and the DHHS began to coordinate dietary recommendations through a joint publication, Dietary Guidelines for Americans, which has been revised every 5 years since then. This publication and numerous government-sponsored reports have been consistent in their recommendation to consume more vegetables and fruit. Over the years, in fact, this recommendation has been elevated in prominence, especially in the Dietary Guidelines edition published in 1995. This background is important to consider because the 5 A Day Program's basic message is not new. By the September 1991 rollout, the essential message had been around for 75 years, with more intensive promotion by the USDA, the DHHS, and other organizations since at least 1980.

National and Local Media Communication Trends

Figure 1 documents the Program's media news and advertising activity from 1991 through 1997. The figure is based on data provided by a professional clipping service and includes both print advertising and print news stories that were entered into a database. However, because the data have occasional gaps and no independent reliability checks were performed, the data should be interpreted more as an index of media placement activity (number of items in the database) than as reliable data about media coverage of the 5 A Day Program. Nevertheless, interpreted in this way, the data demonstrate that substantial placement activity occurred for the 69 months documented. The "launch period" (September 1991-December 1992) showed the greatest amount of activity. It took about 4 months after the rollout for the media to begin publishing large numbers of stories and advertising about 5 A Day, with a major increase occurring in December 1991. Throughout 1992, published material increased sharply to a peak of more than 1,600 items in July 1992. This stayed well above the 1,000 mark through December 1992. Thereafter, activity declined to between 200 to 980 items per month through May 1997, when media clipping ceased.

The taller "spikes" in each year through about 1996, reflect a renewed surge of media promotion activity, usually occurring in September. Following the initial rollout, the Atlanta Constitution and Journal and the North County Times (Oceanside, CA) were the first daily newspapers to report on the Program on September 26, 1991. Through the end of 1991, California print media were the most active in publishing 5 A Day material, perhaps not surprising given the state's reliance on vegetables and fruit in its agricultural economy and its previously established 5 A Day activity.

In the peak year of 1992, all 50 states, Puerto Rico, and the District of Columbia published considerable amounts of material about the Program. The most active state media were in Texas (n = 1,750) and Georgia (n = 1,469), with more than double the promotions of the next most active states: Kansas (n = 676), Oklahoma (n = 593), Florida (n = 527), South Carolina (n = 502), Missouri (n = 440), New Jersey (n = 395), and New York (n = 381). By 1996, the last complete year of media placement data, the rank order of the top 10 most active states was Illinois, California, Kansas, New York, South Dakota, Pennsylvania, Texas, Michigan, Georgia, and Massachusetts.

Table 3 summarizes 5 A Day media placement activity across different regions of the country. The table includes only complete years of data (1992-1996). The most active regions were the Midwest and the South, accounting for about 35 and 34 percent of news and advertising activity, respectively. The Northeast was the next most active, followed by the West, but each of these regions showed about one-half the rate of activity of the Midwest and the South. Midwest activity was fairly consistent throughout the 5 years, while about one-half of the South's activity occurred in 1992. More than one-third of all media activity occurred in 1992, and more than one-half of total activity (52.5%) occurred during 1992-1993.

Figure 1. 5 A Day media and advertising activity September 1991-May 1997
(n = 69 months)

Source: Porter-Novelli

Table 3. 5 A Day media activity by year and region (full years of data only)
Region	1992	1993	1994	1995	1996	Totals	Percent of Totals
West	1,097	1,107	770	641	796	4,411	14.0%
Midwest	2,977	2,043	1,875	2,079	1,991	10,965	34.9%
Northeast	1,358	1,125	925	931	1,016	5,355	17.0%
South	5,360	1,527	1,417	1,285	1,333	10,699	34.0%
Totals	10,792	5,802	4,987	4,936	5,136	31,430	100%
Percent of Totals	34.0%	18.5%	15.9%	15.7%	16.0%	100.0%

Because of the unreliability and incompleteness of the media clipping data, the Evaluation Group conducted an independent analysis of media coverage during the period 1990-1999 to validate trends in news coverage of the 5 A Day message and program. The Lexis-Nexis™ academic database was used, and the search was limited to daily newspapers published during the entire period 1990-1999. This included 12 major U.S. dailies: the New York Times, the San Francisco Chronicle, the Seattle Times, USA Today, the Los Angeles Times, the St. Louis Post-Dispatch, the Washington Post, the San Diego Union-Tribune, the Boston Globe, the St. Petersburg Times, the Louisville Courier-Journal, and the New York Journal of Commerce. The search algorithm included "cancer" in the headline or first paragraph of the story, in combination with "fruit(s)" or "vegetable(s)" in the body text to locate stories related to the 5 A Day message. A second search was conducted using the algorithm "cancer" plus "5 A Day" or "5 A Day for Better Health" in the body text. The results are displayed in Figure 2. Using the first algorithm, the search located 3,041 news stories published in the 12 dailies from 1990 to 1999. This was an average of about 25 stories per month across the newspapers. Using the second algorithm, the Evaluation Group located 135 stories specifically mentioning 5 A Day in conjunction with cancer prevention.

The data demonstrate that the use of the message recommending greater consumption of vegetables and fruit was actually quite high at the beginning of the period (1990), prior to the 5 A Day Program rollout. However, consistent with the clipping data, there was an increase in news coverage of the message from 1991 to 1992, although these data suggest an even greater peak in 1994 compared to the previous 3 years. A smaller proportion of stories mentioned the 5 A Day Program per se, at least using this particular search algorithm. Nevertheless, similar to the clipping database, it appears that news stories about the program peaked in 1992, the year after its initial launch. Although stories leveled off from 1995 through 1996, an additional upswing in coverage of the message occurred in 1997 through 1998. The small number of news stories specifically mentioning 5 A Day is less important than the large number of stories on the message itself, that of consuming at least 5 servings of vegetables and fruit a day.

Figure 2. Major newspaper coverage of cancer, and role of vegetables and fruit, 1990-1999 (12 major market U.S. dailies)


The 5 A Day promotion campaign used a combination of strategies that leveraged advertising from its industry partners and developed relationships with media outlets to generate and inform news stories related to the Program. The media placement data suggest that media relations strategies were less successful after the first 1-2 years of the campaign and that advertising strategies dominated.

Public Response

A great deal of promotional activity drove the 5 A Day message during the 1990s, and major media reported the message in a variety of contexts. To what extent do members of the public know the message, and what are their attitudes toward it? Independent survey data from a variety of sources were examined to answer this question.

Even before the 5 A Day Program, it was clear that a large proportion of the American public understood at least part of the 5 A Day message. In nine national surveys conducted by The Harris Poll, Inc.⁸² between November 1985 and November 1992, Harris asked, "Thinking about your personal diet and nutrition, do you try a lot, try a little, or don't try at all to eat enough fiber from whole grains, cereals, vegetables and fruit?" Results varied little over the 8-year period. Between 52.8 percent and 60.3 percent of U.S. adults reported that they "tried a lot" to eat more fiber from these sources. Only between 8 and 11 percent said they did not try at all.

A second set of national surveys has been conducted biennially since 1991 by the ADA. The most recent was released in January 2000.⁸³ Although 85 percent of Americans say diet and nutrition are important to them personally, about 28 percent say they have made significant changes in their eating patterns to achieve a healthier and more nutritious diet. According to the survey, this number increased by 2 percent since 1997 and is the highest since the survey began in 1991. Moreover, 47 percent of those surveyed said they are doing more to change their diets than they did 2 years ago. About 40 percent of the American public says they know they should eat a more healthful diet but for many reasons have not been able to reach their goal. This number decreased by 2 percent since the 1993 survey. Finally, about 32 percent of Americans say they are unconcerned about making dietary changes, a decrease of 8 percent since the 1997 survey.

A third set of surveys from Federal government sources (FDA, USDA, DHHS), summarized by Guthrie, Derby, and Levy, show that Americans have a high level of awareness about the connection between diet, health, and chronic disease.⁸⁴ For example, in an open-ended question asked from 1982 to 1995 in the FDA's Health and Diet Surveys, the proportion of people mentioning "fats" as linked to heart disease increased from about 20 percent to more than 60 percent. Similarly, about 9 percent spontaneously mentioned the increased consumption of vegetables and fruit as preventive of cancer in 1982. By 1995, this had risen to about 33 percent, an almost fourfold increase. Two surveys in 1994 and 1995 (1994 DHHS/FDA Food Label Use and Nutrition Education Survey [n = 1,945]; 1995 DHHS/FDA Health and Diet Survey-Food Label Use and Nutrition Education Survey Replicate [n = 1,001]) asked specific questions about awareness of the 5 A Day Program. In 1994, 22 percent of those surveyed said they were aware of the Program; 24 percent said they were aware of the Program in 1995.

A fourth set of surveys surrounding the 5 A Day Program was conducted under the auspices of the NCI. A baseline survey conducted in 1991 found that only about 8 percent of Americans thought they should eat at least 5 servings of vegetables and fruit each day.⁸⁵ However, 6 years later (in 1997), a followup survey found that this number had risen to 20 percent.⁸⁶

Dietary Change and Related Outcomes

This section assesses the Program's impact on dietary change and factors that mediate dietary change. The original concept for the 5 A Day Program proposed an evaluation based on two sets of outcomes: (1) changes in two psychosocial factors believed to mediate program effectiveness, specifically the U.S. population's awareness of the 5 A Day Program and knowledge of the 5 A Day dietary recommendation; and (2) changes in the U.S. population's consumption of vegetables and fruit. To collect these data, the evaluation plan included two random-digit-dial, cross-sectional surveys of the U.S. population. Both surveys assessed demographic characteristics, psychosocial factors related to vegetable and fruit consumption, and usual servings of vegetables and fruit. The 1991 baseline survey included 2,834 participants (response rate = 42.8%), ⁸⁷ and the 1997 followup survey included 2,602 participants (response rate = 44.5%).⁸⁸ Staff at the NCI, with the assistance of external consultants, completed analyses of data from both these surveys and other data sources to examine outcomes related to 5 A Day Program effectiveness. The results of these analyses, along with a review of other published reports on U.S. trends in vegetable and fruit consumption, are described on the pages that follow.

Changes in Psychosocial Factors That Mediate Dietary Behavior Change

The assessment of psychosocial factors related to dietary intake is an important and often overlooked component of research and evaluation in public health nutrition. This is because large-scale public health nutrition programs take many years to develop and implement, and thus their influence on dietary behavior will most probably be gradual and cumulative over time. For program evaluation, then, it is useful to examine changes in diet-related psychosocial factors that are believed to mediate intervention effectiveness.^89,90 There is only a small literature on mediating factors for dietary change, but the following factors appear to be most important: knowledge of dietary recommendations,⁹¹ taste preference,⁹² awareness of the benefits of dietary change,⁹³ barriers to change,⁹³ and self-efficacy (confidence that one can perform the desired new behavior).⁹⁴ Many of these constructs were measured in the 5 A Day surveys, with an emphasis on psychosocial factors related specifically to the consumption of vegetables and fruit. Thus, the 5 A Day surveys can be used both to investigate associations of psychosocial factors with vegetable and fruit consumption and to examine how these factors may have changed after the 5 A Day Program was implemented nationwide.

Psychosocial Factors Related to Vegetable and Fruit Intakes

The baseline 1991 5 A Day survey assessed only a small number of psychosocial constructs, and inferences from this survey are limited. The strongest factors predicting vegetable and fruit intakes were knowledge of the dietary recommendation to eat 5 or more servings per day and taste preferences.⁴⁴ In the followup 1997 survey, a more extensive effort was made to include new and revised items to improve the quality of measurement and to assess a broader domain of diet-related psychosocial factors. Table 4 provides results from a comprehensive analysis of the 1997 survey by NCI staff. The single strongest independent predictor of vegetable and fruit intakes was self- efficacy, followed by knowledge of the 5 A Day dietary recommendation and taste preferences. There were weak associations between awareness of the 5 A Day Program, perceived barriers, and perceived norms on the one hand and vegetable and fruit intakes on the other. However, these associations were inconsistent between men and women. There were no associations with perceived benefits, threat, or social support. These analyses suggest that continued monitoring of knowledge of dietary recommendations, self-efficacy for dietary change, and taste preferences can be useful as secondary indicators of 5 A Day Program effectiveness. Improved study and evaluation designs will provide stronger conclusions in the future.

Changes in Psychosocial Factors Related to Vegetable and Fruit Intakes

There is only a little overlap between the diet-related psychosocial factors measured in the 1991 5 A Day survey and those in the 1997 survey. Comparisons are possible based on Program awareness, knowledge of dietary recommendations, and "Stages of Change" toward adopting diets high in vegetables and fruit. Between 1991 and 1997, there were substantial increases in knowledge of the 5 A Day Program, defined as having heard about the Program and correctly identifying it as one that encourages consumption of vegetables and fruit (see Table 5). In 1991, almost no one was aware of the Program. In 1997, 18 percent of Americans were aware of the Program, although when comparisons are made within demographic subgroups, awareness was higher among women, younger people, whites, and better educated people. There also was a substantial increase in the proportion of Americans who knew of the recommendation to eat 5 or more servings of vegetables and fruit each day.

In summary, the 5 A Day message has significantly increased awareness of the importance of eating more vegetables and fruit, in particular among younger and better-educated people.

Table 6 shows the changes in the proportions of the U.S. population in each stage of change toward adopting a diet high in vegetables and fruit. Stages of Change is a heuristic model that describes a sequence of cognitive and behavioral steps in successful behavior change:

• Precontemplation: no recognition of need for or interest in change
  
  
• Contemplation: recognition of need for and thinking about change
  
  
• Preparation: planning for change
  
  
• Action: adopting new behavior
  
  
• Maintenance: continuing practice of new behavior.

Source: National Cancer Institute's 1997 5 A Day Survey, U.S. Adults.

^a Sex, race/ethnicity, age, education, income, marital status, smoking status, BMI, and self-rated health.
^b Demographic characteristics above plus other psychosocial scales in table.
^cScales scored 0,1 (no, yes).
^dScales scored 0-11 (lowest to highest).
^eScale reversed before analysis.
^f p<0.05.
^g p<0.01.
^h p<0.001.

Stages of Change is a key construct of the Transtheoretical Model,⁹⁵ which has been used to design interventions for a wide range of health-related behaviors, including diet.⁹⁶ Between 1991 and 1997, there was a modest 5 percentage point decrease in the proportion of the population in precontemplation, an 11 percentage point decrease in the proportion in maintenance, and a 12 percentage point increase in the proportion in action. Interpretation of these results is not straightforward because the data are from two cross-sectional samples and do not allow direct interpretation of an individual's shift across stages of change. Slight evidence indicates that the proportion of the population that is unaware of, or not at all interested in increasing, its intake of vegetables and fruit has decreased. More speculatively, the decrease in the proportion of the population in maintenance and the increase in those in action stages suggests that about 10 percent of the population has reevaluated its need to increase consumption of vegetables and fruit and is actively engaged in making appropriate dietary behavior changes.

Table 5. Mean percentages of U.S. population aware of the 5 A Day Program and knowing program recommendations in 1991 and 1997
	Program Awareness		Knowledge of Program Recommendation
	1991a (%)c	1997b (%)c	1991a (%)c	1997b (%)c
Total	2	18e	8	19e
Sex
Male	2	14e	4	11e
Female	2	21e	11	27e
Age
18-34	2	22e	7	20e
35-49	3	19e	8	21e
50-64	2	14e	10	18d
65+	0	9e	6	16e
Race/Ethnicity
White	2	19e	8	21e
African American	1	12e	6	13e
Latino	1	10e	6	8
Education (years)
<12	1	16e	5	17e
12	2	16e	7	15e
13+	2	20e	8	22e

Source: National Cancer Institute's 5 A Day Surveys, U.S. Adults.

^a n = 2,834.
^b n = 2,602.
^c Adjusted for age, sex, ethnicity, education, poverty level, and smoking and marital status.
^d vs. 1991, p<0.01.
^e vs. 1991, p<0.0001.

Table 6. Distributions of the U.S. population's Stages of Change for adopting a diet high in vegetables and fruit in 1991 and 1997
	Between 1991 and 1997
Stage of Change	1991a (%)	1997b (%)	Change (%)
Precontemplation	22.1	16.9	-5.2
Contemplation	3.1	2.2	-0.9
Preparation	12.5	18.1	+5.6
Action	35.2	47.2	+12.0
Maintenance	26.7	15.6	-11.1

Source: National Cancer Institute's 5 A Day Surveys, U.S. Adults.

^a Includes all forms, including condiments, candy, chips, and fried food.
^b Mean standard error, adjusted to be representative of the U.S. population during the years of each survey.

Changes in Vegetable and Fruit Intakes

It is extremely difficult to evaluate whether the 5 A Day Program, or indeed whether any mass media-based nutrition campaign, has affected population-level dietary patterns. The reasons are (1) there are no groups not exposed to the 5 A Day campaign that can be used as a comparison to those exposed; (2) there are many other national and local programs to improve dietary patterns, a large proportion of which also include a focus on increasing vegetable and fruit consumption; and (3) low-intensity, public-health oriented dietary intervention programs yield relatively small changes in dietary patterns that may take many years to detect. It is important to consider these issues when interpreting the results for the analyses described on the pages that follow.

The national 5 A Day Program began in 1991 with a national media campaign and promotional activities organized by the PBH; states did not begin intervention programs until 1994. Two strategies are available to evaluate whether the 5 A Day Program has affected dietary behavior during Program implementation: (1) surveillance of U.S. trends in vegetable and fruit consumption and (2) examination of associations at the state-level between the intensity of 5 A Day Program implementation and the magnitude of change in vegetable and fruit consumption.

Surveillance of U.S. Trends in Vegetable and Fruit Intakes

Key outcomes from the two 5 A Day surveys are shown in Table 7. Total consumption of vegetables and fruit (not including french fries) increased by 0.12 servings per day, which was not statistically significant. There were significant increases in vegetable and fruit consumption among Latinos and persons 18-34 years of age. There was a borderline statistically significant difference (p = 0.051) among race/ethnic groups; total consumption decreased among African Americans and increased among Latinos and whites. There were nonstatistically significant larger increases among women compared to men, people with at least a high school degree compared to those not completing high school, and younger compared to older people. Table 7 also includes percentages of the population consuming vegetables and fruit 5 or more times per day in 1991 and 1997. Overall, there was a 4 percentage point increase in the proportion of the population eating 5 or more servings of vegetables and fruit per day, and differences among subgroups were similar to those found for total intake.

There are two additional sources of representative data on U.S. dietary intake that cover the time period corresponding to 5 A Day Program implementation. The CDC maintains the BRFSS Survey, and has published an analysis based on 16 states for the years 1990, 1994, and 1996 (see Table 8).⁹⁷ Vegetable and fruit consumption increased between 1990 and 1994 by 0.14 servings per day (p<0.01), with no further increases between 1994 and 1996. Increases were somewhat larger among women than men. Overall, the percentage of individuals consuming 5 or more servings per day increased by 3.7 percentage points, and increases were larger among women then men. NCI staff have completed a more comprehensive analysis of the BRFSS data based on all 50 states and covering the years 1994, 1996, and 1998 (see Table 9). Between 1994 and 1998, consumption of vegetables and fruit increased by approximately 0.18 servings per day (p<0.001). The proportion of the population eating 5 or more servings of vegetables and fruit per day increased from 21.5 percent to 24.7 percent (p<0.001).

The USDA maintains the CSFII. NCI staff completed an analysis of CSFII data covering the periods 1989-1991 and 1994-1996 (see Table 10).⁹⁸ Total vegetable and fruit consumption increased by 0.3 servings per day among children and by 0.6 servings per day among adults.

Table 7. Mean vegetable and fruit consumption (frequency/day) in 1991 and 1997
	1991 (n = 2,834)		1997 (n = 2,602)
	Freq/day (Mean)a	5+/day (%)a	Freq/day (Mean)a	5+/day (%)a
Total	3.8	23	3.9	26
Sex
Male	3.5	18	3.5	20
Female	4.1	28	4.2	31
Age
18-34	3.5	19	3.8b
35-49	3.7	23	3.8	25
50-64	3.9	24	3.9	25
65+	4.3	33	4.3	33
Race/Ethnicity
White	3.8	23	3.9	25b
African American	4.0	29	3.8	25
Latino	3.6	23	4.0b	31
Education (years)
<12	3.5	19	3.6	21
12	3.6	21	3.7	22
13+	4.0	26	4.1	29

Source: National Cancer Institute's 5 A Day Surveys, U.S. Adults.

^a Adjusted for age, sex, race, ethnicity, education, poverty level, and smoking and marital status.
^b vs. 1991, p<0.05.

Table 8. Mean vegetable and fruit consumption (frequency/day) in 1990, 1994, and 1996a
	1990		1994		1996		Change 1990-1996
	Freq/day se	5+/day (%)	Freq/day se	5+/day (%)	Freq/day se	5+/day (%)	Freq/day se	5+/day %, se
Total	3.27 0.02	19.0	3.41 0.02	22.1	3.40 0.02	22.7	0.13 0.03	3.7, 0.6
Sex
Male	3.10 0.03	16.5	3.19 0.03	18.1	3.20 0.03	19.1	0.10 0.04	2.6, 0.8
Female	3.44 0.03	21.3	3.64 0.03	26.0	3.61 0.02	26.2	0.17 0.04	4.9, 0.8

Source: CDC's Behavioral Risk Factor Surveillance System, 16 states.

^a Adjusted to 1990 pooled age and sex distributions of 16 participating states.
se = standard error

Table 9. Mean of 50 U.S. States' mean vegetable and fruit consumption (frequency/day) in 1994, 1996, and 19981
	1994	1996	1998
Total	3.77a	3.85b	3.95c
Fruit	0.77a	0.80b	0.81b
Juice	0.70a	0.72a	0.75a
Vegetable	1.48a	1.50a	1.55b
Salad	0.48a	0.48a	0.49b
Potatoes	0.37	0.38	0.379
Percent 5+/day	21.5a	23.2b	24.7c

Source: CDC's Behavioral Risk Factor Surveillance System.

¹ Mean, adjusted for smoking, obesity, agricultural sales, age, sex, race, and marital status.
^{a, b, c} Means without same superscript differ significantly across years (p<0.05).

Table 10. Mean vegetable and fruit consumption (servings/day) in 1989-1991 and 1994-1996
	Fruita		Vegetablesa		Total Vegetables and Fruita
	1989-1991	1994-1996	1989-1991	1994-1996	1989-1991	1994-1996
Total Age   Total (2+ yrs)	1.3+ 0.03b	1.5 + 0.03	3.2 + 0.03	3.4 + 0.04	4.5 + 0.06	4.9 + 0.05
2-19 yrs	1.3 + 0.06	1.6 + 0.05	2.6 + 0.07	2.7 + 0.06	4.0 + 0.09	4.3 + 0.08
20+ yrs	1.3 + 0.04	1.5 + 0.03	3.4 + 0.05	3.7 + 0.04	4.6 + 0.06	5.2 + 0.05

Source: USDA's Continuing Survey of Food Intake by Individuals (CSFII).

^a Includes all forms, including condiments, candy, chips, and french fries.
^b Mean standard error, adjusted to be representative of the U.S. population during the years of each survey.

Taken together, these three large survey programs suggest small increases both in mean vegetable and fruit consumption and in the proportion of individuals consuming vegetables and fruit 5 or more times per day in the time period during implementation of the 5 A Day Program. Increases in vegetable and fruit consumption were substantially smaller based on the 5 A Day and the BRFSS surveys as compared to the CSFII survey, which may have several explanations. First, both the 5 A Day and the BRFSS surveys used a short food frequency questionnaire method that captured only the frequency of consuming vegetables and fruit, without information on portion size, and included only those forms of vegetables and fruit that are targeted by the 5 A Day intervention. In contrast, the CSFII survey used multiple dietary recalls that captured details on all forms of vegetables and fruit, including those found in mixed dishes and condiments. Second, the analyses of both the 5 A Day and the BRFSS surveys were designed to examine change over time, by adjusting the data from each survey to a common distribution of population demographic characteristics; the analyses of the CSFII data are based on the population demographic characteristics at the time of each survey. Consequently, some of the increase observed in the CSFII survey is probably due to the increasing age and educational level and decreasing smoking rates in the U.S. population, as all of these characteristics are associated with higher vegetable and fruit consumption.

In summary, there has been a slow and steady increase in vegetable and fruit consumption in the United States during the implementation of the 5 A Day Program. Possible inferences from these results on the effectiveness of the 5 A Day Program are limited. The possibility cannot be ruled out that without the 5 A Day Program there would have been substantial decreases in vegetable and fruit consumption, paralleling the enormous increase in obesity over the same time period. Also, it is likely that other economic and cultural factors are influencing dietary behavior change in the United States. Nevertheless, these results are consistent with the inference that the 5 A Day Program has contributed to the continuous small increases in vegetable and fruit consumption over the past decade.

Intensity of 5 A Day Program Implementation and Changes in Vegetable and Fruit Intakes

The original 5 A Day concept included a process evaluation based on biannual state surveys of program activities. Data collected between 1995 and 1998 were used to create a state-level "implementation index," a summary scale with a range of 1 to 4 based on: (a) state health agency 5 A Day expenditures, (b) state health agency 5 A Day staff hours, (c) 5 A Day print materials used, and (d) 5 A Day ancillary materials used. In addition, a newspaper clipping service collected data on the number of articles that specifically mentioned the 5 A Day Program between 1991 and 1997. It was therefore possible to examine whether there were associations at the state level between the intensity of program implementation and the annual number of newspaper articles, and the magnitude of change in vegetable and fruit consumption.

NCI staff completed two sets of analyses. The first set examined whether the implementation index was correlated with change in mean state-level vegetable and fruit consumption during the period 1994 through 1998. This analysis asked the question, "Were there larger increases in consumption of vegetables and fruit in the states that more fully implemented 5 A Day Program activities?" These analyses found no associations between implementation intensity and change in vegetable and fruit consumption. A second, parallel analysis examined the associations between the number of newspaper articles and changes in consumption but found none. Additional analyses examined these associations in sex and age subgroups, but also found none.

There are several limitations to these analyses based on implementation intensity. The data used to calculate the implementation index were designed for process evaluation only, and the entire domain of activities describing state-level program implementation was not captured. The intensity of program implementation was not large; even the most populous states had no more than one FTE of staff time devoted to the 5 A Day Program. Newspapers are only a small portion of the media; television is not captured using this approach. Finally, many of the state programs targeted school-aged children, but the only available measure of state-level vegetable and fruit consumption is based on adults.

In summary, these findings suggest that more intensive 5 A Day Program implementation and more newspaper coverage were not associated with larger increases in vegetable and fruit consumption. However, given the limitations in the data available for these analyses, these conclusions are not strong.

Randomized Trials and Other Studies

The NCI funded nine formal, randomized experimental trials to investigate the effects of behavioral programs on fruit, juice, and vegetable consumption. These studies, taken as a whole, were successfully implemented and substantially increased understanding of how to motivate healthy eating practices under a variety of settings and within diverse populations. Four of the nine studies tested interventions in school settings. The remaining five studies focused on adults in worksites, churches, and WIC clinics. Tables 11 and 12 present a summary of the findings from eight of the nine studies; one has not yet been completed.

In Table 11, the data from the youth studies provide evidence that multicomponent school-based interventions can improve the health behaviors of elementary school children from diverse ethnic and social backgrounds. The studies employed state-of-the-art intervention techniques and evaluation designs. The interventions included components such as classroom curricula; modifications to the school food service; direct marketing of fruits, juices, and vegetables to children at lunch and at school; involvement of the vegetable and fruit industry; parent involvement; and community-based social marketing strategies. Across the studies, the net differences between the treatment and control groups at the first posttest ranged from 0.20 to 1.68 servings of vegetables and fruit per day. The average effect size was 0.68 servings per day; most programs found statistically significant increases. The strongest intervention effects were found on daily fruit consumption, possibly because of the higher palatability of fruits. For two of the studies (Georgia and Louisiana), significant intervention effects were observed after the first year of intervention, but these effects diminished in the last year of intervention. The Alabama study indicates that effects can be maintained without continued intensive intervention, although the differences were smaller at followup. Similar effects were found in an evaluation of the California Children's 5 A Day Power Play Program, which was not funded under this NCI mechanism. In this study, fourth and fifth grade students exposed to a school and community intervention ate 0.40 more servings of vegetables and fruit as compared to students in the control schools.

Table 11. Youth school studies: Effects of 5 A Day intervention studies on daily servings of fruit, juice, and vegetables
Project Title	Target	Intervention Strategy	Design	Main Outcome Measurement	Net Change Between Treatment and Control (Bold = P<.05)
5-A-Day Power Plus99 Minnesota	4th-5th graders 48% White 25% Asian 19% African American 6% Hispanic 60% Free/reduced lunch	2 years of intervention Classroom curricula Parent involvement School food service Industry involvement	10 School pairs Randomized Pre-posttest/control group School unit analysis N final cohort = 441	24 hour recall:   -total servings    -per 1,000 kcal Lunch direct obs.   -total servings   -per 1,000 kcal	PT: FV = .58;   F = .62; V = .02 PT: FV = .41;   F = .36; V = .05 PT: FV = .47;   F = .30; V = .16 PT: FV = .83;   F = .72; V = .23
Gimmie 5100 Georgia	4th-5th graders 85% White 5% African American	2 years of intervention Classroom curricula Parent involvement Food industry	8 school pairs Randomized Pre-posttest/control group School unit analysis N final cohort = 1,253	7 day recall	PT: FV = .20;   F = .12; V = .08
Alabama High 5101 Alabama	4th-5th graders 83% White 16% African American 1% Other Median Income @ $45,000	2 years of intervention Classroom curricula Taught by program staff Parent involvement School food service	14 school pairs Randomized Pre-posttest/control group Delay intervention School unit analysis N final cohort = 1,426	7 days of 24-hour recall 5 a day FV score Lunch direct obs.	PT: FV = 1.68;   1F = .88; V = .69 FU: FV = .99   F = .56; V = .35 PT: FV = 1.46   F = .77; V = .50 FU: FV = .85   F = .50; V = .23 PT: FV = 0   F = .05; V = -.03 FU: FV = .09   F = -.02; V = -.03
Gimmie 5: A Fresh Nutrition Concept102 Louisiana	9-12th graders 84% White 4% African American 9% Hispanic 3% Other	3 years of intervention Food service marketing Student workshops Parent component	6 school pairs Randomized pre-posttest/control group School unit analysis N final cohort = 1,911	Daily consumption of FJV; single item self report	PT: FV = .30

FV = Daily servings of fruit, juice, and vegetables. F = Daily servings of fruit. V = Daily servings of vegetables.
PT = Posttest (intake measured 3-12 months after intervention).
FU = Followup (intake measured 3-12 months after intervention).

The school results, taken together, offer compelling evidence of program effectiveness for elementary school children across sex, race, and economic subgroups. The implications for research and practice include the need to: (1) fund the dissemination of elementary school interventions that work, perhaps by combining "best practices" from across programs; (2) continue research on why children eat the foods they do and explore other interventions in addition to those used in these research projects with promise for impact on vegetable and fruit consumption; (3) develop and evaluate innovative programs for middle and high schools; (4) develop and evaluate a stand-alone school food-service module that combines best practices from elementary schools; (5) develop and evaluate school and community policy approaches that increase the availability of vegetables and fruit in schools (e.g., fruit juice in vending machines, breakfast carts with vegetables and fruit, etc.); and (6) develop methods for innovative, effective, widespread teacher training.

The adult studies (see Table 12) also provide evidence that multicomponent interventions can improve the health behaviors of individuals in different cultural and social settings. These studies rigorously tested innovative intervention strategies among individuals at worksites, churches, and WIC programs. At the first posttest, each study found statistically significant increases in daily vegetable and fruit consumption as compared to controls. These increases ranged from 0.20 to 0.85 servings per day, with an average effect size of 0.48 servings per day. The strongest intervention effects were found for daily fruit consumption.

A common element among the studies was the inclusion of peer educators or intervention channels that also targeted the social environment. The Arizona and Massachusetts studies compared traditional worksite strategies to worksite plus peer education (Arizona) or worksite plus family education (Massachusetts) and found that family- or peer-led interventions were significantly more successful in increasing vegetable and fruit consumption than were interventions focusing exclusively on the worksite. These findings support the underlying hypothesis that dietary behaviors occur in a social context and demonstrate how the effectiveness of nutritional interventions can be enhanced when they take into account an individual's social context—including home, church, and peer networks at the worksite. The findings underscore the important role that social systems, including family members, coworkers, and church members, have to play in determining the climate of health behavior and how such social systems can assist in improving eating habits.

The adult results, taken together, also offer compelling evidence of program effectiveness across sex, race, and economic subgroups. The implications for research and practice include the need to: (1) fund dissemination of worksite-based programs, with a particular emphasis on programs targeting aspects of the workers' social contexts, perhaps by combining "best practices" across programs; (2) continue research on factors mediating the effectiveness of interventions in increasing vegetable and fruit intake; (3) develop and evaluate interventions in other settings, including point-of-purchase settings such as grocery stores and restaurants; and (4) develop and evaluate interventions that incorporate messages about vegetables and fruit with other behavioral risk factors such as physical activity or tobacco control.

Table 12. Adult studies: Effects of 5 A Day intervention studies on daily servings of fruit, juice, and vegetables
Project Title	Target	Intervention Strategy	Design	Main Outcome Measurement	Net Change Between Treatment and Control (bold = P<.05)
Healthier Eating for the Over-looked Worker103 Arizona	75% Male lower income 46% White 41% Hispanic 6% African American 6% < high school 34 % high school 49%>high school	1.8 years of intervention Peer education plus general 5 A Day compared to general 5 A Day alone	93 randomized work cliques Pre-posttest/control group Control group 6 month followup Clique unit of analysis N final cohort = 695	24-hour food recall 7-item 30-day food frequency	PT: FV = .77   F = .41; V = .26; J = .10 FU: FV = .41   F= .06; V = .24; J = .11 PT; FV = .46   F = .25; V = .19; J = .01 FU: FV = -.04   F = .03; V = -.08; J = .07
Black Churches for Better Health104 North Carolina	73% Female 65% < $20,000 98% African American 37% < high school 34% high school 29% > high school	20 months of intervention Multicomponent intervention: tailored print materials, direct, education, lay health advisors, community coalitions, church activities, grocery	5 matched randomized county pairs 49 churches Pre-posttested/control group County unit of analysis N final cohort = 2,519	7-item 30-day food frequency	PT: FV = .85   F = .66; V = .19
Maryland WIC 5-A-Day Promotion Program94 Maryland	100% Female lower income 53% African American 43% White 19% < high school 41% high school 37% > high school	6 months of intervention Nutrition sessions by peer leaders Print materials and visual reminders Direct mail	16 WIC sites randomized Pre-posttest/control group 1 year followup WIC site unit of analysis N final cohort = 695	7-item 30-day food frequency	PT: FV = .43 FU: FV = .74
Treatwell 5-A-Day105 Massachusetts	84% Female 59% White 23% Hispanic 18% African American 20% = high school 36% some college, vocational 42% college	19.5 months of multicomponent intervention: worker participation, individual and environmental changes, family component	22 worksites randomized into minimal intervention (8), worksite plus family (7), worksite only (7) Pre-posttest/control group Worksite unit of analysis N final survey = 1,306	7-item 30-day food frequency	PT: FV, Worksite + Family = .50 PT: FV, Worksite FV = .20

FV = Daily servings of fruit, juice, and vegetables. F = Daily servings of fruit. V = Daily servings of vegetables. J = Daily servings of juice.
PT = Posttest (intake measured 3-12 months after intervention).
FU = Followup (intake measured 3-12 months after intervention).