In North Carolina, approximately 77% of adults do not consume the recommended daily intake of 5 or more servings of fruits and vegetables and 26% do not engage in leisure time physical activity (3,4). These statistics underscore the prevalence of unhealthy behaviors and chronic conditions in this US region, yet provide little information about the motivations and perceptions of residents relative to these factors. Especially important is the paucity of data on peoples' qualitative descriptions of their dietary behaviors in the context of their environment. This is of particular interest given that lifestyle choices, including those relating to diet, are complex decisions affected by the interaction between people (attitudes, thoughts, behaviors, perceptions) and their external social and physical environments (7-9).

Historically, most assessments of dietary and other lifestyle behaviors and attitudes have been conducted using survey methods. However, because of their systematic structure, surveys have been less able to contextualize individuals' experiences and perceptions and do not facilitate easy assessment of motivational factors and contexts underpinning peoples' behaviors (10). Nonetheless, they have been useful for identifying patterns of behavior and facilitating comparisons of patterns across environments (10). However, interventions based on survey results, primarily individual-focused education strategies, have been limited in effecting behavior change and improving health outcomes (11). Researchers suggest that the absence of corresponding environmental changes may be responsible for this outcome (11). In fact, despite numerous preventive efforts regarding dietary behavior and chronic diseases, obesity and other chronic conditions continue to increase in prevalence. This change indicates that the mere identification of unhealthy behaviors and development of educational strategies to change them may not be enough. Researchers often state that people's behaviors are contextual and must therefore be understood and addressed within their environment (7,8,9,11-13). Yet, few studies have investigated individuals' perceptions of the relationship between their environment and diet and physical activity behaviors (11). Until recently, no unifying conceptual framework existed to evaluate these relationships. However, a work group of researchers has produced such a framework. The work group summarizes its effort as the beginning phase in clarifying the relationship between the environment and lifestyle behaviors and strongly encourages further inquiries on "the extent of environmental influence and how [environment and lifestyle] affect different individuals" (11).

An approach that has proved useful in both identifying and contextualizing lifestyle behaviors is formative research. This unique multimethod, exploratory approach combines social science techniques such as interviews, focus groups, and other qualitative methods with quantitative methods such as surveys (14). When these techniques are used to gather data for informing research conduct and intervention design, the term formative assessment is used. Formative research has been used primarily by intervention researchers in community-based studies and has guided the development of several interventions (13,15,16). A unique strength of this approach is that it provides the basis for collection of relevant new data to guide research direction, research discourse, and data interpretation (13,15-17). Drawing on a large study to assess community and social determinants of chronic disease risk and outcomes among North Carolina adults, we used formative research to report specifically on findings relative to fruit and vegetable intake.

The study proceeded in 2 phases: a telephone survey followed by a formative assessment phase. For the formative assessment, the research protocol included focus groups, semistructured individual interviews, photographs, and cursory community mappings.

Study participants were residents of North Carolina who had previously consented to be part of a unique statewide practice-based research network called the North Carolina Family Medicine Research Network (NC-FM-RN) (18). The NC-FM-RN was initiated in 2001 and currently includes a cohort of North Carolina residents previously recruited through a network of 25 family practice settings across the state. The cohort was enriched with new participants in 2004 and 2005. The sample for this study included people who were part of the cohort in 2001 and 2004. They were recruited from 22 of the 25 sites. Further details on the structure, recruitment methods, and outcomes of the NC-FM-RN are available elsewhere (18). All the methods and the protocols for this study and the NC-FM-RN were approved by the University of North Carolina at Chapel Hill Medical Institutional Review Board.

At the study's outset in 2004-2005, previously enrolled participants from 22 of the 25 practices in the NC-FM-RN were sent a letter that described the study, its objectives, what participants could expect, and reminded participants that they had previously consented to be contacted for future studies. Consenting participants were scheduled for the telephone survey, which included a 30-minute assessment of demographic, health, and social factors, and health attitudes, behaviors, and beliefs related to chronic diseases. A total of 2,479 people from the cohort participated in the telephone survey, resulting in a 59.5% response rate. All measures in the survey were self-reported. Race was identified as non-Hispanic black, non-Hispanic white, or other (inclusive of all other races). Comorbidity was measured as reported diagnosis of 1 or more of 18 chronic conditions. Some of the conditions assessed were heart disease, hypertension, diabetes, and depression. Comorbidity scores were calculated by summing the number of reported diagnoses. Body mass index (BMI) was calculated from self-reported weight and height, and was reported as weight in kilograms/(height in meters)². It was categorized according to Centers for Disease Control and Prevention guidelines (underweight, <18.5; normal weight, 18.5-24.9; overweight, 25.0-29.9; and obese, ≥30.0) (19). The survey also included the following 3 questions to assess participants' perception of the quality, variety, and affordability of fruits and vegetables in their environment:

The response options for the first 2 questions ranged from excellent to not applicable and for the third question, very affordable to not applicable.

Participants were eligible to participate in focus group discussions if they had completed the telephone survey. Six of the 22 NC-FM-RN practices were chosen as recruitment sites, and patients from these sites were invited to participate first by letter and then by telephone calls. A total of 84 people were recruited, and 21 (25%) attended the discussion sessions. Seven focus groups, averaging 3 participants and lasting 1.5 hours, were conducted in 6 locations. Participants received a gift of $20. Sessions were arranged, facilitated, and audiotaped with participants' consent by 2 trained moderators (B.S., L.F.C.) and subsequently transcribed, coded, and analyzed (B.S., K.R.M.) using ATLAS.ti version 5.0 software (Scientific Software Development, GmbH, Berlin, Germany). Nutrition-specific thematic content and coding analyses were further conducted (J.E.A.B.). Portions of the transcripts that addressed participants' reported consumption patterns of fresh fruits and vegetables and their perceptions of environmental influences on their nutrition behaviors were reviewed for emergent themes and compared across groups. Participants responded to the question, "Do you think about getting fruit and vegetables in your diet?"

To broaden the range of perspectives, semistructured individual interviews were conducted with community members not demographically represented in the focus groups. These individuals were recruited from the same list used for the focus group participants. A total of 11 participants who were African American men, African American women, or white men were intentionally recruited to participate in telephone-administered interviews. This addition increased the formative assessment sample to 32.

Before each focus group session, a cursory geographic assessment of each of the 6 localities was conducted. Research team members drove through the downtown sections of localities on the day of the session and identified physical assets that could serve as resources for or constraints to lifestyle behavior change. Relevant assets included the availability, type, and proximity of grocery stores, farm stands, restaurants, and convenience stores. This endeavor helped moderators to contextualize the behaviors and perceptions reported by the participants during the sessions.

Participants were asked to identify and photograph factors in their environment that they perceived demonstrated a relationship between their community and health. They were each given a disposable camera and a photo-diary with instructions on how to document their thoughts about their photographs. They were also instructed that the purpose of this activity was to facilitate discussion and that the activity was not a requirement for participation in the discussions. Submitted photographs were coded as representative of barriers or facilitators if participants showed them to the group in their responses to nutrition questions or if the photographs noticeably related to nutrition.

To identify demographic characteristics and group differences, survey results were analyzed using univariate and bivariate statistics. Differences in perceptions of affordability, accessibility, and variety of fruits and vegetables in neighborhood stores were tested by categories of race, BMI, socioeconomic status (SES) as determined by income and poverty rate by census block group, education, and chronic disease status using χ² test of differences at the α = .05 level. The poverty rate by census block group variable was created by using 2000 census data, which indicated that 12.3% of the North Carolina population lives below the poverty level. Focus group transcripts were content-analyzed for prevalent perceptions related to fruit and vegetable intake (20,21). Identified perceptions were then categorized by using the conceptual framework proposed by Booth and colleagues (11) (Figure, Table 1).

The framework includes individual (intrinsic), lifestyle-enabling, and external (extrinsic) environmental factors. Pictorially (Figure), the framework consists of 7 concentric rings categorized from the innermost to the outermost ring as follows: psychobiologic core, cultural, social, enablers of choice, behavioral settings, proximal leverage points, and distal leverage points. Each ring consists of specific factors that are separate from those in other rings (further described in Table 1). The individual's environment includes the psychobiologic core and cultural and social segments. The psychobiologic core consists of the individual's current physical health status and encompasses factors such as genes, instinctive and conditioned behaviors, and experiential learning. The cultural segment encompasses values, traits, and beliefs acquired from the social and cultural environment. The social segment covers how people perceive themselves and their social roles, and includes acquired beliefs and values (11). Positioned between the individual and the external environments are the enablers of choice. These are the barriers and facilitators to change that are proximal to the individual and include factors that facilitate interaction between the individual and the external environment to promote or hinder the lifestyle behavior of interest. The external environment includes the remaining 3 outermost rings in the framework. The first of these 3, labeled behavioral settings, are the contexts, both physical and social, in which lifestyle behaviors occur. Proximal leverage points are "controllers of the structure and feature of the microenvironment that affect the choices related to the behavior of interest" (11). Finally, distal leverage points reflect macro-level influences on behavior and address systemwide factors such as the health care system, the information industry, the food industry, and the government. Application of this model to the qualitative transcripts allowed us to categorize participants' perceptions as either internal or external environmental influences on fruit and vegetable intake. In this report, the terms intrinsic and extrinsic are used to describe the individual (personal) and external (physical/social) environments, respectively.

Participants responded to 3 survey questions assessing food quality, variety, and affordability (Table 3). BMI and counts of comorbid conditions were associated with significant differences in perceptions about quality of fruits and vegetables available at the grocery store where participants regularly shopped. The number of comorbid conditions was also significantly related to perceptions of variety. For affordability, we found significant differences by income, education, BMI, poverty rate by census block group, and comorbid conditions (Table 3). Age was significant for all 3 questions.

Transcripts revealed a rich dataset of perceptions related to factors perceived to affect intake of fruits and vegetables. Findings from the community mappings helped to contextualize participants' perceptions. Photographs shown and discussed by participants during the sessions included images of farms, restaurants, kitchen spaces, convenience stores, gardens, buffet foods, and condiments, and generally supported participant perceptions. We used the framework by Booth and colleagues to categorize perceptions on the basis of related environmental factors highlighted in the framework and by the influence (barrier or facilitator) of the perception on dietary behavior (Table 4). Most perceptions were categorized as either a barrier or a facilitator. However, some had dual influences.

The following intrinsic factors were identified as barriers: food preferences, fatigue of taste buds for certain foods, life stresses, lack of forethought in meal planning, current personal health status, aging, and perceived impact of food on current chronic disease status. Intrinsic factors perceived as facilitators were the presence of chronic disease, lifetime experience related to intake of fruits and vegetables, preferences for certain fruits and vegetables, and personal or spousal health status. At the extrinsic level, participants reported the following factors as facilitators: availability of home gardens, low cost of foods at farm stands, and childhood exposure to fruits and vegetables. Perceived barriers included contradictory media messages related to nutrition and health outcomes, worksite food options, food availability, and food cost at grocery stores. Finally, participants reported the following as factors perceived to have an interactive effect: concerns about food safety and perceptions about the interaction between chronic disease status and social and environmental influences on behavior and health. For example, participants highlighted the interactions between physical fatigue due to changing work schedules or shifts and stresses resulting from trying to manage fatigue, work schedule, and personal dietary intake at home. Some participants perceived chronic disease status as a facilitator, whereas to others it was a barrier.

One potential limitation of this study is the small sample used for the formative assessment. Researchers, however, indicate that, regardless of qualitative methodology used (focus group or key informant interviewing), with adequate representation, a sample of 30 individuals is enough to uncover the perceptions of the majority of a population (26,27). Other limitations of the study are that intake of fruits and vegetables, nutrition behavior, and weight and height data (for BMI) were not directly measured. Hence, correlations and causality cannot be verified for these variables.

Participants' perceptions of intrinsic and extrinsic influences on intake of fruits and vegetables are consistent with those of other studies (23,28-30) but specifically indicate that in this sample, behaviors and attitudes toward fruits and vegetables are influenced by both intrinsic and extrinsic environmental factors. Furthermore, they provide grounds for reconsidering the focus of approaches to dietary interventions to see how salient they are in addressing the perceived needs for (or barriers to) behavior change in this population. Many interventions are based on findings from surveys that tend to focus on individuals' behaviors with less emphasis on context. Our results suggest that a complex web of factors and perceptions underpins participants' nutrition behaviors. Overall, the results suggest that interventions should use multifactorial approaches that acknowledge the contexts of individuals and their environmental limitations and should provide options that people can use to achieve healthy lifestyle habits. Interestingly, most of the perceptions described by the participants of this study were more intrinsic than extrinsic. This finding suggests that individually focused interventions would be useful in helping them with chronic disease risk management. Family practice settings, from which all the participants of this study were recruited, may be an ideal place to start because they can serve as important sources of information for health promotion and disease prevention (22,31-33). A recent study confirms that individually tailored messages can significantly affect the nutrition behaviors of people who visit family practice settings (33). However, researchers also indicate that the "time has come for interventionists (whether public health or primary care) to look more closely at context — the risk-laden conditions within communities — rather than limit the focus to individual risk factors" (34). Thus, the emphasis should not be limited to individually focused efforts but expanded to include linking individuals and family practice settings with available community-based resources. This effort is feasible for North Carolina; already several community-based programs promote healthy nutrition habits. One notable program is the Eat Smart Move More North Carolina statewide initiative (www.eatsmartmovemorenc.com). This program targets all North Carolinians and has offerings designed for a multitude of settings and audiences. These offerings can be individually tailored, and when executed properly, allow individuals to access resources that promote risk reduction by increasing healthy nutrition and physical activity practices.

The findings of this study indicate that many unfavorable intrinsic and some extrinsic factors are perceived to affect the intake of fruits and vegetables of this sample of North Carolinians. The perceptions evidenced are of concern because they are associated with behaviors that increase chronic disease risk. Options that would facilitate increased fruit and vegetable intake are needed, and family practice settings and community-based programs may be useful places to begin.