Health Education Research Advance Access originally published online on September 18, 2006
Health Education Research 2007 22(3):425-437; doi:10.1093/her/cyl092
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Evaluation of a nutrition education intervention for women residents of Washington, DC, public housing communities
1 Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, 610 North Wolfe Street, Baltimore, MD 21205, USA
2 Department of Health, Behavior, and Society, Johns Hopkins Bloomberg School of Public Health, 624 North Broadway, Baltimore, MD 21205, USA
3 Department of Oncology, Johns Hopkins School of Medicine, 600 North Broadway, Baltimore, MD 21205, USA
4 Department of Biostatistics, Johns Hopkins Bloomberg School of Public Health, 610 North Wolfe Street, Baltimore, MD 21205, USA
* Correspondence to: A. C. Klassen. E-mail: aklassen{at}jhsph.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsConclusionsConflict of interest statementAcknowledgementsReferences |
|---|
We designed, implemented and evaluated an educational intervention to increase fruit and vegetable consumption among urban African-American women. Women aged 20–50 years (n = 212) from 11 public housing communities participated in seven 90-min classes with a professional nutritionist. Our prospective pre- and post-test design, with 4-month follow-up, assessed the relationship between attendance and dietary change, using three 24-hour recalls per time point. Mean change in average daily dietary values for fruits and vegetables, calories and percent calories from fat (post-test versus pre-test, follow-up versus pre-test) was compared by class attendance, to evaluate the impact of class attendance on dietary change. Attendance varied from zero (35%) to five to seven classes (42%). Baseline dietary recalls showed average daily consumption of 3.05 servings of fruits and vegetables, 2416 calories and 35.8% calories from fat. No improvements in fruit and vegetable consumption, but statistically significant decreases in total calories and percent calories from fat, were seen at both endpoints. Women attending five to seven classes had the greatest dietary improvements, averaging, at post-test and follow-up, respectively, 246.2 and 324.5 fewer calories and 3.08 and 2.97% fewer calories from fat. Results suggest that, for some residents of low-resource communities, small group interventions are popular, effective vehicles for nutrition education.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsConclusionsConflict of interest statementAcknowledgementsReferences |
|---|
Nutritional factors may contribute to 20–60% of cancers worldwide and to one-third of deaths from cancer in Western countries [1]. Epidemiologic studies indicate that consumption of fruits and vegetables in significant amounts is associated with reduced risk of many types of cancer and cardiovascular disease [2, 3]. Eating five servings of fruits and vegetables daily is recommended to reduce risk of these diseases [3]. In addition, a large body of research has investigated and demonstrated the potential role of anti-oxidant nutrients in the prevention of cancer, cardiovascular disease, cataract and age-related macular degeneration [4]. Anti-oxidants, including carotenoids, vitamin E, vitamin C and the trace mineral selenium, are abundant in many fruits and vegetables [5]. Although findings from recent large prospective studies and meta-analyses are less supportive of a blanket protective role for total fruit and vegetable consumption, certain phytochemicals may still have strong protective effects, such as lycopene-containing cruciferous and allium vegetables [6].
National data on dietary patterns demonstrate that the US African-American population experiences diet-related cancer and cardiovascular disease risk equivalent to or greater than the white population [7]. Intake of dietary fiber, fruits and vegetables is lower in African-Americans than in whites and intake of nitrite-cured and smoked foods is higher [8]. Among low-income women of both white and black races, some assessments have shown low intakes of fiber, vitamin A and calcium, along with high intakes of fat, saturated fat, cholesterol and salt [9]. In addition, the prevalence of obesity is greater among African-American females than in white females [10].
Minority and low-income populations are important targets for interventions to improve diet, in order to reduce excess chronic disease burden in these groups [11]. One important pathway to dietary improvement is through nutrition education. However, low levels of economic and social resources in disadvantaged communities may make it difficult for these populations to assimilate, and make use of, media-based information from national nutritional campaigns such as 5 a Day and the Food Guide Pyramid. As stated by Stables et al. [12], because of the intricate relationship between food intake patterns and culture, ‘nutrition professionals need to target specific demographic subgroups with tailored interventions to move all Americans toward achievement of dietary guidelines for vegetable and fruit consumption’.
In conjunction with the national 5-a-Day communication campaign, a range of interventions targeting nutritional knowledge, attitudes and behaviors among both children and adults have been undertaken [13]. Consistently, the adult studies have demonstrated the importance of building interventions based on the social context, by recognizing family, church or worksite networks as important influences on nutritional behaviors. However, one remaining question is whether these types of interventions are translatable into populations such as the chronically unemployed or impoverished, whose everyday social networks may be more difficult to identify and utilize for public health initiatives.
Consistent with the national data for African-American adults [7–10], a dietary pattern associated with greater cancer incidence and mortality is characteristic of the Washington, DC, African-American, low-income community [14]. Our previous research among this population [15] revealed that in most households, women served as the primary nutritional decision makers, with at least one meal per day being prepared for the entire household to eat together. We found a desire among women to improve their nutrition-related skills, including menu planning, budgeting, shopping and meal preparation. Equally important, our formative work revealed that women had significant psychosocial needs related to the successful management of family nutrition with limited resources. Therefore, intervening on knowledge, skills and self-efficacy appeared central to successfully changing dietary behaviors.
The goals of this research were to (i) develop a nutrition intervention to increase fruit and vegetable consumption tailored specifically for African-American women living in public housing communities in Washington, DC, and (ii) evaluate the program by measuring change in dietary knowledge, attitudes and behaviors among program participants. In this paper, we report on program design and implementation, as well as process evaluation measures of program attendance. We then evaluate program effectiveness by analyzing the relationship between program attendance and our primary outcome, dietary behavior change.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsConclusionsConflict of interest statementAcknowledgementsReferences |
|---|
Intervention population
The study population consisted of African-American women of child-rearing age (20–50), residing in selected Washington, DC, public housing communities. With the support of the District of Columbia Housing Authority, the elected officers of the resident councils in each public housing community were utilized as liaisons for our nutrition education program in their community. Residents were recruited through announcements at monthly resident council meetings, posters and door-to-door contact. All interested women were screened for eligibility. We limited eligibility to African-American women whose parents and grandparents were born in the United States, and excluded other residents of African ancestry, such as Afro-Caribbeans, because of potential differences in dietary habits [16]. Additional exclusion criteria included physical inability to participate in all aspects of the intervention activities, disclosure of significant drug or alcohol use, serious medical conditions, medically prescribed or pre-packaged diets, pregnancy or lactation. Participants in general nutritional programs such as Women, Infants and Children (WIC) were not excluded.
Intervention design
Our goal was to design a program that could be easily replicated in other low-income communities. Therefore, our intervention was designed to be relatively brief and low cost in terms of time and materials. Table I provides an overview of program elements and time points for our prospective pre- and post-test design, with 4-month follow-up.
|
Based on our review of similar interventions [13], we designed a program of six 90-min sessions conducted twice a week for 3 weeks, followed with one 90-min ‘booster’ session held 6 weeks later. In reviewing the literature on nutritional interventions [13], we found that outcome evaluation time points varied with the type, intensity and duration of the intervention. Given the relatively brief duration of our intervention, we operationalized post-test dietary change as change evident at Week 4, immediately following the formal program, and more sustained dietary change, and possible evidence of durable program effect, as dietary change at follow-up, 20 weeks after enrollment.
We recruited and trained a female African-American registered dietician, who also had significant experience in community-based nutrition education, to lead the seven-session, small group, nutrition program, conducted in the community center kitchen of each participating housing complex. Results of extensive formative work, including focus group interviews, provided insight into potential participants' preferences regarding educational content, session length and scheduling and characteristics of teachers and fellow learners.
Intervention content
Our social–ecological theoretical framework [17] recognized multiple levels of influences on food behaviors. Therefore, the goal of the intervention was to enhance knowledge and skills needed to combat the many structural and social barriers to healthy eating within the participants' lives. Program content and activities combined nutrition education and food-related skill development. To build self-efficacy for nutrition-related problem solving [18], psychosocial tools such as food-related parenting strategies, goal setting and planning and seeking alternative solutions to problems were emphasized. Additionally, in order to address the sense of isolation many of these single parents of young children reported [15], women were encouraged to contact their class partners outside of class to share transportation for shopping, cook together or offer support. In recognizing that the true target of the intervention was the entire household, discussions focused on strategies for involving family members in the program goals and ‘winning them over’ to dietary change. The style of the sessions was interactive, focusing on demonstration of techniques by the nutrition educator, followed by direct practice of new skills by participants. Psychosocial techniques to encourage interaction among classmates included games, contests and other ‘ice breakers’. Participants were paired at the first class, and pairings were used for both in-class and outside-of-class activities.
Content was tailored specifically to young and middle-aged women who had grown up in low-resource urban food environments. For example, many participants had limited experience with different types of cookware, kitchen tools such as measuring cups and spoons, kitchen sanitation and food safety, multiday household food planning or written recipes. Skills were practiced in the same environment where they would need to be sustained. For example, shopping tours took place in local supermarkets, and ‘homework’ involved repeating classroom tasks at home, such as receiving money to purchase and prepare foods for class.
The role of pictures in the intervention
Pictorial information delivery was an important theory-based element of the intervention, contributing to four components: engagement, comprehension, information retention and behavior change. Below, we summarize the ways in which our use of pictures as an intervention tool is supported by extensive research, which we have recently reviewed [19].
Most fundamentally, pictures helped to ‘engage’ participants, and keep their attention focused on the information being presented during the sessions. Research demonstrates that people are more likely to pay attention to information that is accompanied by explanatory pictures. Furthermore, because culturally tailored pictures help make information personally salient, we used a professional artist to create project materials featuring urban African-American families with positive attitudes toward fruit and vegetable consumption.
Second, pictures facilitate ‘comprehension’ of information beyond what can be understood with words alone. The special contribution of pictures is to integrate information and to help people make inferences which extend beyond what was presented in words. This was an important component of the training, in that we sought to change behavior at home, where the women had to make decisions about their families' menus by generalizing the principles underlying specific examples used in class. Additionally, research supports the importance of linking pictures to clear simple language in spoken or written text. Instructors carefully avoided technical language when linking pictures to nutrition concepts. The pictures were also designed to communicate key points with minimal reliance on written language. For example, recipes for class cooking projects consisted of pictures showing each step, such as measuring cups or spoons filled to the correct level. As a result, participants could follow the recipes without relying on text.
Third, pictures can help participants ‘retain’ what they learned, which was crucial if messages were to influence behavior outside of training. Studies show that text accompanied by pictures is better remembered than text alone. This so-called ‘pictorial superiority effect’ [20] is especially strong when people see pictures with an oral explanation and later see the same pictures to remind them of what they heard. We used this principle in several ways.
During the instructor's oral presentations, she showed related pictures and pointed to relevant elements in the pictures as they were discussed. This ensured that participants linked pictures with information. At the end of each session, participants added copies of the pictures used in that session to binders they were compiling, and were encouraged to review them at home. In addition, they received laminated place mats that included pictures and simple text summarizing the main messages for each session. These place mats, designed to be attractive as well as informative, were to be used across the following week with family members at mealtimes to discuss new foods or dietary changes being introduced. These place mats prompted participants to educate family members and create family support for recommended nutrition behaviors.
Fourth, pictures can increase the likelihood that people will ‘want to carry out’ recommended health behaviors. Research in marketing indicates that the emotional response to the picture affects whether it increases or decreases a target behavior. Thus, we used appealing pictures with positive family-oriented associations in conjunction with explanations of the need for increasing fruit and vegetable consumption. Additional information and visual examples of these materials are available [21].
Data collection
Dietary data
Dietary data were collected at three time points: at baseline, immediately after intervention classes and 4 months after the intervention. Computer-assisted interviewer-administered 24-hour dietary recalls were used to assess dietary intake, using the Nutritional Data Systems for Research (NDS-R) version 4.03.31 (University of Minnesota Nutritional Coordinating Center, Minneapolis, MN). All persons collecting nutritional data were trained, certified and re-certified annually to meet NDS-R standards. Three non-consecutive recalls (two weekdays, one weekend) were collected per subject at each time point. The NDS system records all foods consumed, and therefore, it is possible to analyze data in multiple ways, from specific micronutrients to foods by brand or food groups.
Non-dietary data
At recruitment, data collection staff administered a brief face-to-face questionnaire to assess participant eligibility. At each time point (baseline, post-intervention and follow-up), enrolled participants completed the face-to-face dietary recalls. After nutritional data were collected, the face-to-face interviewer-administered questionnaires were collected. The baseline interview measured an extensive range of psychosocial influences in the participants' lives, including sociodemographics, family composition and psychological status, as well as knowledge, attitudes and practices related to food preparation and consumption. Post-intervention and follow-up interviews measured reactions to the intervention, as well as knowledge and attitudes about food. Measures were chosen from the literature, including our previous surveys among urban African-American women [22–25]. All questionnaires were pilot-tested with community members not involved in the intervention.
At the initial training, the data collectors were trained to measure height and weight. Height was measured once at the initial interview and the weight was taken at the initial and at the consecutive interviews throughout the study. Height and weights were measured without shoes. To measure height, the participant was asked to stand straight against the wall, and a flat board was placed on the head and the top was marked with a pencil, the length was measured using a metal tape and recorded in inches. Weight was measured using a portable digital scale, which was calibrated to zero before weighing each participant. The weight was recorded in pounds. The same weighing scale was used throughout the study.
In addition to the nine 24-hour recalls and these four interviews described above, two additional data collection activities took place. During each session, an observational checklist was completed by research staff not directly involved in teaching, recording process evaluation data on participants (attendance, mood, behaviors), classroom (functioning of equipment, interruptions such as noise or visitors, flow of activities) and instruction (responsiveness to participant mood or questions, completeness and quality of session content). Community questionnaires were completed by resident council leaders, whose residential tenure typically made them knowledgeable sources for characteristics of the community food environment, such as accessibility of grocery stores and home gardens, as well as community characteristics such as size, resident tenure and other current or prior nutrition activities.
To maintain the integrity of the intervention activities and ensure confidentiality of data, none of the data collection staff was involved in the intervention. Each participant was modestly compensated for time spent in data collection, to encourage participation and obtain the most complete data possible; compensation ranged from $5–$25, depending on activity. No compensation was offered for attendance at intervention sessions because one research goal was to elicit attendance patterns generalizable to non-research use of the intervention. The project was approved by the institutional review board of the Johns Hopkins Bloomberg School of Public Health. Participants provided written informed consent before determination of eligibility and before enrollment in the intervention itself.
Data analysis
Descriptive data on the participating communities were obtained from the community questionnaire completed by community leadership about their housing complexes. We supplemented these data with Census tract–level descriptive data from the 2000 Census on median household composition and socioeconomic status (http://www.census.gov). This information is reported in Table II.
|
Data from participant interviews were used to examine baseline psychosocial characteristics among all participants, as well as differences in these psychosocial characteristics by class and data collection participation rates. The first column in Table III displays univariate frequencies for psychosocial characteristics for all participants in the analysis. The second set of columns compares psychosocial characteristics of respondents by class attendance status; we used the chi-square statistic to test for significant variation. The third set of columns similarly compares psychosocial characteristics by data collection status.
|
Measures reported in Table III from the eligibility and baseline interview include respondent age, years of formal education, current employment status, number of persons and number of children in the household, residential tenure, place of birth, driver's license, participation as a voter in the 2000 presidential election and smoking status. Body mass index was calculated from interviewer-measured height and weight, using the standard formula ((weight in pounds/(height in inches2)) x 703 (www.cdc.gov/nccdphp/dnpa/bmi/adult_BMI/about_adult_BMI.htm). Recent depressive symptoms were measured with a brief version of the Center for Epidemiologic Studies Depression (CES-D) inventory [26], and literacy was measured by respondent score on the Rapid Estimate of Adult Literacy in Medicine [27], an interviewer-administered 60-word pronunciation test.
Table IV describes the baseline, post-intervention and follow-up dietary patterns among participants, and compares these dietary patterns among respondents of differing attendance levels. Dietary data were summarized using descriptive statistics: means, medians, standard errors and ranges. Primary outcome variables included servings of fruits and vegetables, total calories and percent calories from fat.
|
For fruits and vegetables, we combined servings from 11 NDS categories (citrus juice, fruit juices excluding citrus juice, citrus fruit, fruits excluding citrus fruit, dark green vegetables, deep yellow vegetables, tomatoes, legumes, other vegetables, fried vegetables and vegetable juice) but did not include fried fruits, fruit-based savory snacks, white potatoes, fried potatoes or vegetable-based savory snacks. The NDS system captures data on all foods consumed; therefore, partial servings (such as a slice of lettuce in a sandwich) contributed to overall totals.
For each outcome variable at each time point (baseline, post-class and follow-up), the average was taken over the 24-hour recalls available for each person. Because some respondents did not provide three recalls at each time point, we use statistical weighting techniques to minimize any effect introduced by the low reliability of fewer recalls.
The mean across women at each time point was calculated using a weighted average of the mean recalls per woman where the weight was the square root of the number of recalls used in computing the mean (i.e. analytic weights were used) [28]. To calculate differences between post-class and baseline, or follow-up and baseline, weights were based on the product of the number of recalls at each time point divided by the sum. To test for differences from baseline to post-class and follow-up, we used weighted linear regression, with P values from t tests used for inference. Alpha of 0.05 was used for determining statistical significance. Larger P values could represent either the lack of an association or insufficient power for the association of interest. We used SAS, release 9.1 (SAS Institute, Inc, Cary, NC), for dietary analyses and SPSS, release 12.0 (SPSS Corporation, Chicago, IL), for all other analyses.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsConclusionsConflict of interest statementAcknowledgementsReferences |
|---|
After a 16-month development phase, recruitment, data collection and intervention activities began in September 2001, and across a 28-month time period, we conducted 18 waves of the 20-week intervention and follow-up in 11 public housing communities, located in the southeast and northeast areas of the District of Columbia. The 18 intervention cohorts ranged in size from 5 to 17 women (average size = 10).
Of 234 women initially expressing interest, 221 signed consent for eligibility screening, 217 were eligible and 212 women signed consent to participate in the intervention. Of these, 187 women began the data collection process by providing at least one dietary recall prior to the first class, and are included in the analysis described below.
Table II provides information on the 11 communities in which the study's public housing units were located and the number of housing units, and therefore potential program audience, in each complex. Fifty-five percent of the 187 participants lived in the southeast areas of DC, and the remaining 45% lived in the northeast areas. Fifty-nine percent of participants lived in complexes with gardening, and 42% lived within five blocks of a supermarket. The 2000 Census information for the Census tracts in which these housing complexes were located shows that these communities all are moderately or extremely disadvantaged. Median values for Census tract characteristics were 97% African-American, 75% female-headed families with children under the age of 18 years, 58% high school graduates among adults, 48% household car ownership and median household income of $21 130 annually.
Table III displays psychosocial characteristics of participants, and bivariate relationships between these characteristics and the level of program participation. From the univariate distributions, we see that the majority of participants are over the age of 30 years, and that almost two-thirds have completed high school or equivalent. Only 17% of participants hold a job, and most live in households with children. There is good residential stability, and, perhaps because the majority of participants are lifetime residents of Washington, DC, few have ever had a driver's license. Most participate in voting. There is some evidence of physical and mental health burden among the participants: 61% currently smoke, 49% are obese (BMI 
30) and 29% report moderate or severe levels of depressive symptoms. One-fifth of participants read at below a sixth grade level.
Thirty-five percent of enrolled participants attended no classes, and 42% attended most or all of the seven sessions. Attendance was higher among older women, those not currently holding jobs, those born in the District of Columbia (DC), non-smokers and obese women. Class attendance was a strong predictor of data collection participation. Although rates fell among all groups across the three waves of data collection, attrition was highest among women who did not attend classes.
At baseline, 80% of women in this analysis completed three recalls and the interview, with few psychosocial differences between completers and non-completers. Only place of birth was significantly associated with differences; women born in DC were significantly more likely to complete all baseline data collection.
However, at post-intervention data collection, several characteristics differentiated those who completed all aspects of data collection from those who did not. Differences in data collection participation mirror class attendance differences. Those more likely to complete data collection include older women, longtime residents and those born in DC and those without CES-D symptoms at baseline.
At follow-up, data collection participation was greater among older women, those with residential stability and DC birth, voters, non-smokers, obese women and those with higher reading levels. Finally, complete data collection was best among older women, longtime residents and those born in DC, voters and those with higher reading levels.
Table IV provides information on dietary patterns among participants. At baseline, respondents consumed on average 3.05 servings of fruits and vegetables and 2416 calories daily, with 35.8% of calories coming from fat.
In the aggregate, participants' average servings of fruits and vegetables did not change significantly between baseline and post-test or between baseline and follow-up. When we separate women according to number of classes attended, we see slight evidence of a positive trend associated with class attendance, with high attenders having an average increase of 0.26 servings, partial attenders increasing by an average of 0.17 servings and non-attenders having a negative change of –0.13 servings. These changes are not statistically significant. At follow-up, there is no statistically significant change among attenders. However, interestingly, there is a significant decrease of –0.20 servings in average fruit and vegetable consumption among non-attenders.
Although fruit and vegetable consumption does not appear to have been significantly changed by class participation, two other indicators of healthy diet do appear to have improved among class attenders. Average total calories decreased among all participants between baseline and post-test by 225 calories and between baseline and follow-up by almost 300 calories. Percent calories from fat decreased by almost 3% and continued to show this decrease at follow-up.
In terms of class participation differences in these changes, the evidence is mixed, but suggests greater improvement for attenders than non-attenders, especially in percent calories from fat. Non-attenders show a statistically significant average decrease in one of the four values (total calories at post-test), and partial attenders show statistically significant decreases in two values (calories at post-test and percent calories from fat at follow-up). However, only those attending five or more sessions show statistically significant changes in all four values. Impressively, they ate an average of 251 fewer calories at post-test and 331 fewer at follow-up. They also showed a sustained decrease in calories from fat (3% at both time points). We tested results for cohort, housing community and seasonality effects, but these were not significant.
|
Conclusions |
|---|
|
TopAbstractIntroductionMethodsResultsConclusionsConflict of interest statementAcknowledgementsReferences |
|---|
As stated in Introduction, our research had two goals. The first was to develop a nutrition education program that engaged low-income women, and sustained their participation, in competition with the myriad of other demands on their time and attention. In this regard, we were reasonably successful. Many women enrolled willingly in the program, reported that they enjoyed the activities and classes and returned to class as frequently as they could manage.
However, many women who expressed initial interest did not attend all classes. To explore this, we conducted focus group interviews post-program with groups of both high and low attenders. Some women cited barriers as simple as walking from their town house unit to the community center in extremely hot or cold weather. More fundamentally, although our formative work suggested that weekday classes were most likely to be attended, women who initially enrolled subsequently missed classes for mandatory activities such as job training.
Between initial planning and completion of our project, free time for low-income women to attend discretionary daytime activities was greatly reduced, due in part to new programs such as ‘Welfare to Work’. Although our goal was to reach all families in these communities, one unanticipated effect was that our potential audiences became limited to women who were truly unemployable due to significant chronic mental or physical illness. Possibly, reasons contributing to joblessness also presented barriers to program success. Moreover, a policy-based transition from low-income to mixed-income public housing changed the composition of these communities during this time. Finally, other community activities competed with our program for residents' attention, often offering more immediate incentives such as payment or job credits. These historical events are typical of the challenges to conducting community-based behavioral intervention trials.
The second goal of our research was to enable participants to improve their dietary patterns, and sustain changes over time. We see evidence that women who attended all the classes made dietary changes, and sustained those changes for 4 months after classes ended. Interestingly, the dietary change that was the primary intervention focus, increased fruit and vegetable consumption, was not seen as often in participants as two other dietary improvements, fat reduction and calorie reduction. There are several possible reasons for this.
The program presented healthy eating as a unified concept, including controlling portion size, interpreting food labels and evaluating food content and using recommended food guidelines to plan healthy meals. Despite our caveats at enrollment, in our post-intervention focus groups, some women gave desired weight loss as one motivation for enrolling [29]. This suggests that messages about healthy eating may have been more salient than specific messages about fruits and vegetables.
We found other examples of differences between our program goals and the participants' goals. Participants revealed that one incentive for participation was to achieve documented nutrition-related competencies useful when job seeking in the food service industry. Given our university affiliation, our certificate of participation had to be carefully negotiated to offer an accurate yet desirable representation of program ‘graduation’.
The extensive high-quality data collected were a strength of our design, but also a potential limitation. Despite incentive payments for data collection, some women were lost to follow-up. Therefore, our results may be biased toward women we could re-locate. Conventional standards for reliability in dietary measurement call for three 24-hour recalls, including one weekend. However, these recommendations are based on the typical workweek. In our population, we found that weekend patterns did not vary significantly from the weekday, most likely because these women did not routinely leave home on weekdays and remain home on weekends. However, we did find a substantial amount of variation across each woman's recalls, regardless of day of the week. This suggests that even three recalls may not reliably capture all the variations in low-income urban diets.
Our results are difficult to compare to 5-a-Day interventions [13] for several reasons. First, most interventions took place over a longer period, with more resources, among more socially advantaged groups, such as church members or the employed. Our intervention, by design, sought to engage the most disadvantaged populations, and test the feasibility and impact of a relatively low-cost intervention, which could possibly be sustained by communities with minimal investment. Second, most interventions relied primarily on food frequency questionnaires, which are more vulnerable to measurement bias than the 24-hour recall [30]. Finally, due to these measurement limitations, other fruit and vegetable interventions do not typically report other dietary changes, making it difficult to compare our caloric and fat decreases [13].
Despite these differences, we found that our best attenders did show an increase in fruit and vegetable consumption of a quarter serving per day, which, perhaps due to small sample size, did not achieve statistical significance. Given that most 5-a-Day interventions saw increases ranging from 0.20 to 0.85 after multiyear, multicomponent efforts [13], we believe this further supports the potential value of this intervention.
Further analyses are being conducted to consider how subgroups of women responded differently to program elements, and whether there was variation in outcomes by psychosocial characteristics. However, these aggregate results demonstrate that our intervention produced dietary improvement in attenders, and should be considered as one possible model for nutrition intervention in low-resource households and communities.
|
Conflict of interest statement |
|---|
|
TopAbstractIntroductionMethodsResultsConclusionsConflict of interest statementAcknowledgementsReferences |
|---|
None declared.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsConclusionsConflict of interest statementAcknowledgementsReferences |
|---|
This research project was supported by American Cancer Society grant TURPG-00-294-01-PBP. The authors would also like to acknowledge the enthusiastic support and assistance of the residents and staff of the participating housing communities, as well as the data collection and nutrition education staff. Copies of visual materials used in the nutrition education sessions are available for health education use from the corresponding author on request.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsConclusionsConflict of interest statementAcknowledgementsReferences |
|---|
1. Doll R. The lessons of life: keynote address to the nutrition and cancer conference. Cancer Res (1992) 52:2024s–9s.
2. Hung HC, Joshipura KJ, Jiang R, et al. Fruit and vegetable intake and risk of major chronic disease. J Natl Cancer Inst (2004) 96:1577–84.
3. Krauss RM, Eckel RH, Howard B, et al. AHA Dietary Guidelines: revision 2000: a statement for healthcare professionals from the Nutrition Committee of the American Heart Association. Circulation (2000) 102:2284–99.
4. Mayne ST. Antioxidant nutrients and chronic disease: use of biomarkers of exposure and oxidative stress status in epidemiologic research. J Nutr (2003) 133:933S–40S.
5. Willett WC. Diet and cancer: an evolving picture. J Am Med Assoc (2005) 293:233–4.
6. McCullough ML, Giovannucci EL. Diet and cancer prevention. Oncogene (2004) 23:6349–64.[CrossRef][ISI][Medline]
7. American Cancer Society. Cancer Facts and Figures for African Americans 2005–2006. (2005) Atlanta, GA: American Cancer Society.
8. Krebs-Smith SM, Cleveland LE, Ballard-Barbash R, et al. Characterizing food intake patterns of American adults. Am J Clin Nutr (1997) 65:1264S–8S.
9. Breslow RA, Subar AF, Patterson BH, et al. Trends in food intake: the 1987 and 1992 National Health Interview Surveys. Nutr Cancer (1997) 28:86–92.[ISI][Medline]
10. Hedley AA, Ogden CL, Johnson CL, et al. Overweight and obesity among US children, adolescents, and adults, 1999–2002. J Am Med Assoc (2004) 291:2847–50.
11. US Department of Health and Human Services. Healthy People 2010. 2nd Ed. With Understanding and Improving Health and Objectives for Improving Health. 2 Vols. (2000) Washington, DC: US Government Printing Office.
12. Stables GJ, Subar AF, Patterson BH, et al. Changes in vegetable and fruit consumption and awareness among US adults: results of the 1991 and 1997 5-a-Day for Better Health Program surveys. J Am Diet Assoc (2002) 102:809–17.[CrossRef][ISI][Medline]
13. 5-A-Day Evaluation Group. 5-A-Day for Better Health Program Evaluation Report (2000) Bethesda, MD: The National Cancer Institute.
14. Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance System Survey Data (2002) Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention.
15. Shankar S, Klassen AC. Influences on fruit and vegetable procurement and consumption among urban African-American public housing residents, and potential strategies for intervention. Fam Econ Nutr Rev (2001) 13:33–45.
16. Lancaster KJ, Watts SO, Dixon LB. Dietary intake and risk of coronary heart disease among ethnic subgroups of black Americans. J Nutr (2006) 136:446–51.
17. Smedley BD, Syme SL. Promoting Health: Intervention Strategies from Social and Behavioral Research (2000) Washington, DC: National Academy Press.
18. Houts PS, Shankar S, Klassen AC, et al. A problem-solving approach to nutrition education and counseling. J Nutr Educ Behav (2006) 38:254–8.[CrossRef][ISI][Medline]
19. Houts P, Doak C, Doak L, et al. The role of pictures in improving health communication: a review of research on attention, comprehension, recall, and adherence. Patient Educ Couns (2006) 61:173–90.[CrossRef][ISI][Medline]
20. Levie WH. Research on pictures: a guide to the literature. In: The Psychology of Illustration—Willows DM, Houghton HS, eds. (1987) vol. 1. New York: Springer-Verlag. 1–50.
21. Houts PS, Shankar S, Klassen AC, et al. Use of pictures to facilitate nutrition education for low income African American women. J Nutr Educ Behav. in press.
22. Klassen AC, Smith ALM, Meissner HI, et al. If we gave away mammograms, who would get them? A neighborhood evaluation of a no cost breast cancer screening program. Prev Med (2002) 34:13–21.[CrossRef][ISI][Medline]
23. Klassen AC, Juon HS, Alberg AJ, et al. Opportunities for oral cancer screening among older African American women. Prev Med (2003) 37:499–506.[CrossRef][ISI][Medline]
24. Greene AL, Torio CM, Klassen AC. Well-screened: measuring sustained mammography use by African-American women. J Community Health (2005) 30:235–51.[CrossRef][ISI][Medline]
25. Papas M, Klassen AC. Pain and discomfort associated with mammography among urban, low-income African American women. J Community Health (2005) 30:251–65.
26. Radloff LS. The CES-D scale: a self-report depression scale for research in the general population. Appl Psychol Meas (1977) 1:385–401.[CrossRef]
27. Davis TC, Long SW, Jackson RH, et al. Rapid estimate of adult literacy in medicine: a shortened screening instrument. Fam Med (1993) 25:391–5.[Medline]
28. Neter J, Kutner M, Nachtsheim C, et al. Applied Linear Statistical Models (1996) 4th edn. Chicago: Irvin.
29. David EM, Clark JM, Carrese JA, et al. Racial and socioeconomic differences in the weight-loss experiences of obese women. In: Am J Public Health (2005) 95:1539–43.
30. Willett W. Nutritional Epidemiology (1998) 2nd edn. New York: Oxford University Press.
Received on October 22, 2005; accepted on July 17, 2006
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||