Disclosure: The authors declared no conflict of interest.
Supermarket discounts of low-energy density foods: Effects on purchasing, food intake, and body weight
Article first published online: 5 JUL 2013
Copyright © 2013 The Obesity Society
Volume 21, Issue 12, pages E542–E548, December 2013
How to Cite
Geliebter, A., Ang, I. H., Bernales-Korins, M., Hernandez, D., Ochner, C. N., Ungredda, T., Miller, R. and Kolbe, L. (2013), Supermarket discounts of low-energy density foods: Effects on purchasing, food intake, and body weight. Obesity, 21: E542–E548. doi: 10.1002/oby.20484
Funding agencies: This study was funded by a seed grant from the Robert Wood Johnson Foundation and the Columbia University Institute for Social and Economic Research and Policy (ISERP).
- Issue published online: 3 DEC 2013
- Article first published online: 5 JUL 2013
- Accepted manuscript online: 17 APR 2013 01:07PM EST
- Manuscript Accepted: 22 MAR 2013
- Manuscript Received: 6 JUN 2012
To assess the effects of a 50% discount on low-energy density (ED) fruits and vegetables (F&V), bottled water, and diet sodas on shoppers' purchasing, food intake, and body weight.
Design and Methods
A randomized, controlled trial was conducted at two Manhattan supermarkets, in which a 4-week baseline period (no discounts) preceded an 8-week intervention period (50% discount), and a 4-week follow-up period (no discounts). Twenty-four hour dietary recall, as well as body weight and body composition measures were obtained every 4 weeks. Participants (n = 47, 33f; 14m) were overweight and obese (BMI ≥ 25) shoppers.
Purchasing of F&V during intervention was greater in the discount group than in the control group (P < 0.0001). Purchasing of these items by the discount group relative to the control group during follow-up was reduced from intervention (P = 0.002), but still remained higher than during baseline (P = 0.01), indicating a partially sustained effect. Intake of F&V increased from baseline to intervention in the discount group relative to the control group (P = 0.037) and was sustained during follow-up. Body weight change did not differ significantly between groups, although post hoc analysis indicated a change within the discount group (−1.1 kg, P = 0.006) but not within the control group.
Discounts of low-ED F&V led to increased purchasing and intake of those foods.
The rise in obesity rates over the past several decades likely derives from environmental changes that encourage increased energy intake and decreased energy expenditure [1-3]. Treatment and prevention strategies focusing on the behavior of the individual, typically yield only modest results [2-4], and appear inadequate in the face of the obesogenic food environment. In individually based weight loss treatments, more than 33% of an individual's weight is regained within 1-year post-treatment, and 90% is regained within 5 years . Pharmacological approaches also have limited effectiveness, with FDA-approved medication producing only 5-10% weight loss while the medication is taken [6, 7]. Invasive treatments such as bariatric surgery produce greater and longer lasting weight loss but have limited application. Of the 22 million Americans clinically eligible, only 0.4% undergo bariatric surgery . Given these challenges, attention has shifted from individual-based approaches toward interventions at the environmental level [9-13]. An increasing number of studies are testing this approach [14-20], and to date, several such interventions have shown promising results, including weight loss [19, 20].
This study employed an environmental economic intervention to reduce the price of foods low in energy density (ED), defined as kilocalories per gram (kcal g−1) . Individuals generally tend to consume a constant volume of food, regardless of macronutrient or energy content [22-24]. Increasing the intake of low-ED foods has been shown to concurrently decrease the intake of high-ED foods [25, 26]. Substituting low-ED foods for high-ED foods (e.g., replacing a cookie with a fruit) could theoretically reduce total energy intake and lead to weight loss [25, 27, 28]. Many households in the US have a limited budget for food  and spend only a small fraction of their income on fruits and vegetables [25, 26].
Several environmental interventions have led to greater purchasing of low-ED foods . A 50% price reduction applied to cafeteria fruits and vegetables, or to low-fat (≤3 g fat) snacks from a vending machine significantly increased the purchasing of these items [15-18]. Additionally, increased purchasing of healthy foods resulted from a supermarket discount and correlated with decreased purchasing of high-fat foods . Those studies, however, examined purchasing of the items only, and did not measure food intake and body weight.
This study is the first to examine not only purchasing but food intake and body weight changes following a price reduction intervention. We predicted greater purchasing and intake of low-ED items during the intervention period by the discount group than by the nondiscount control group, which might lead to decreased body weight and body fat.
The objective of this study was to examine the effects of discounting low-ED fruits and vegetables, bottled water, and diet sodas by 50% on purchasing, food intake, body weight, and body composition (fat and lean mass). The discounts were implemented during an 8-week intervention period that was preceded by a 4-week baseline period (no discounts) and followed by a 4-week period (no discounts). Twenty-four hour dietary recalls, as well as body weight and body composition measures were obtained every 4 weeks (Figure 1). The study was a randomized, controlled trial that was registered with ClinicalTrials.gov.
Overweight (BMI ≥ 25 kg m−2) and obese (BMI ≥ 30 kg m−2) shoppers were recruited from two D'Agostino Supermarkets (at 91st St. and at 110th St.), about one mile apart in Manhattan, NYC. Shoppers were recruited in a rolling fashion from within the store and via study advertisements printed on their store receipts. There were no apparent differences in the demographic characteristics of shoppers or in food item availability or display between the two stores. Median annual family incomes were similar for the surrounding census tracts of both stores: $49,750 for the 91st St store and $46,154 for the 110th St store . Although shoppers do not always live near the supermarket at which they shop, we found that of the 38 of the 47 participants, for whom we had home addresses, 32 lived within 10 min walking time (0.5 mile) and 6 within 25 min walking time (1.2 mile) of their supermarket.
Of the 104 individuals screened, 67 qualified and were enrolled in the study (Figure 2). Of the participants, 10 completed less than half of the study and were considered dropouts. Dropouts did not differ significantly from completers in age, initial BMI, or gender. Dropout rates did differ by study condition (discount: 7%: control: 30%), which could be due to participants dropping out when told just prior to the start of the intervention period that they were randomized into the control group, and therefore would not receive the discounts. Those who completed at least half of the intervention period were included in the analyses. Ten participants were excluded poststudy due to noncompliance or incomplete purchasing data.
The final number of participants included in the analysis was 47 (33 women, 14 men): 28 in the discount group, 19 in the control group (Figure 2). The mean BMI was 30.2 kg m−2 (SD = 4.5; range = 25.0-40.3 kg m−2), and the mean age was 37.5 years (SD = 13; range = 21 to 65 years). Of these, 56% were Caucasian, 19% African American, 9% Asian, 13% Hispanic, 2% selected “Other,” and 2% provided no response. The baseline characteristics of the discount and control groups did not differ and are presented in Table 1 as means and standard deviations (SD). All participants signed hospital approved IRB consent forms.
|Characteristics||Discount (n = 28)||Control (n = 19)|
|Age, mean (SD), y||37.1 (12.4)||38.1 (14.2)|
|BMI, mean (SD), kg m−2||30.2 (4.3)||30.2 (4.9)|
|Female, no. (%)||20 (71)||13 (68)|
|Male, no. (%)||8 (29)||6 (32)|
A structured phone interview was used to determine eligibility. Candidates were screened for BMI ≥ 25 kg m−2 based on reported height and weight, subsequently verified during the initial consultation. They were included only if they reported purchasing more than 50% of food at their D'Agostino supermarket, and then signed an agreement that they would shop for all groceries exclusively at that supermarket for the duration of the study. Participants could dine out or have take-out for dinner up to two times weekly. To maximize the correspondence between food purchasing and intake, participants had to be the primary food shopper in their households, shopping for up to one other person, not counting children 6 years old or younger. They were excluded if they reported significant medical or psychiatric conditions. Participants were also requested not to start a weight loss program during the study.
Participants were first stratified by gender and then randomized within each store, using a randomizer program (random.org) into two groups: one receiving a 50% discount on low-ED foods and beverages, and the other not receiving a discount. Participants were blind to their group assignments during the 4-week baseline period, and were informed of their group assignment just before the 8-week intervention period began.
There were three sequential time periods for both groups: a 4-week baseline period, an 8-week intervention period, and a 4-week follow-up period, when prices reverted to baseline for the discount group. The discount group received a 50% discount on selected fruits, vegetables, and non-caloric beverages during the 8-week intervention period, while the control group received no discounts during this period. Assessments of key outcome measures (dietary intake, body weight, and body composition) were obtained at procedural time points when participants met with research staff in their supermarket every 4 weeks, at the start (0 week), end of the baseline period (4 week), mid-intervention (8 week), end of intervention (12 week), and end of follow-up (16 week). Procedures for the assessment of key outcome measures were conducted in a private office of the supermarket to ensure confidentiality. Purchasing data were collected continuously from 0 to 16 week via the scan cards. Discounted foods included common fruits and vegetables that were low-ED and excluded high-ED fruits and vegetables, such as dried or canned fruits, avocados, and potatoes. Noncaloric beverages (bottled water and diet sodas) were also discounted.
Price reduction intervention
This study utilized the existing D'Agostino Rewards scan cards that most customers already used. All participants turned in their prior D'Agostino Rewards scan cards and received new cards to track purchasing for the duration of the study and to implement discounts for the discount group during the intervention period. When making payment for store items, customers presented the scan card, which had a barcode linked to their unique account. Participants were randomized after the 4-week baseline period to either the discount or control group. A list of the discounted fruits, vegetables, and noncaloric beverages were provided to the discount group, and the same list was also provided to the control group as a list of healthy food items. Price reductions were automatically applied via the scan cards belonging to those in the discount group for the entire 8-week intervention period.
Type, quantity, and cost of items purchased were referenced by a given scan card number and separated by study period. Mean weekly purchasing data of discounted and nondiscounted items were collected and analyzed in blocks by study period (baseline, intervention, follow-up). Because the scan card was programmed to track specific purchasing data of the discounted items only, purchasing data were not available for total food, sugar sweetened beverages, and less healthful foods and beverages. Purchasing data were also unavailable for the dropouts.
Dietary intake data
A 24-h recall of food intake was administered in person by a trained research assistant at each procedural time point. The 24-h recall was used to obtain a 1-day sample of fruit and vegetable intake, beverage intake, and total energy intake during each time period. Fruit and vegetable intake were expressed in grams (g) as well as serving sizes. A serving size represents 80 g of fruits and vegetables, which is the internationally recognized standard [34, 35]. Noncaloric beverage intake was expressed in grams (g) and energy intake in kilocalories (kcal). Noncaloric beverage intake was based on the sum of commercially bottled water, noncaloric sodas, and tap water. Energy intake (kcal) from caloric beverages was based on the sum of kcal from caloric sodas, alcoholic beverages, and presweetened beverages, such as sweet tea, mochas and lemonades. Intake data from 24-h dietary recalls were analyzed using FoodWorks (The Nutrition Company, Long Valley, NJ).
Body weight and composition
A Tanita BF-680W scale was used to measure body weight (kg) and to obtain percentage body fat (% fat) by bioimpedance analysis (BIA). Measurements were performed in person at each procedural time point by appointment within the supermarket. A Seca 214 stadiometer was used to measure height at the initial consultation to calculate BMI.
Mixed-design ANOVA with repeated measures was used to analyze the main effects over time and the interaction effect between group (discount vs. control) and time on outcome measures. When significant (P < 0.05), or near significant differences (P = 0.05-0.07) were found, post hoc tests were employed to analyze specific changes. That is to say, when there was an overall effect of time, changes over time within each group were examined, and when there was a significant interaction between group and time, comparisons between groups for more specific time periods were made. Outcome measures were: weekly purchasing of discounted fruits and vegetables (gross $ or nondiscounted prices) and discounted noncaloric beverages (gross $); intake of fruits and vegetables (g; serving sizes), and of noncaloric beverages (g); energy intake from caloric beverages (kcal) and total energy intake (kcal); body weight (kg), body fat (kg), and fat-free mass (kg).
After the main analyses were completed, age, gender, race, initial BMI, and season of the year when the participant began the study were entered as covariates. None of these differed between groups, and the results did not change after adjusting for the covariates. Results for all main analyses and for significant secondary post hoc analyses of outcome measures are given by F and P values, and effect sizes. Means and standard deviations (SD) are also shown for significant results. Effect sizes are expressed as Cohen's d, with d = 0.2 considered a small effect size, d = 0.5, medium, and d = 0.8, large . Only the P values are shown for nonsignificant secondary post hoc analyses. Two-tailed P ≤ 0.05 were required for statistical significance. The Statistical Package for the Social Sciences (SPSS version 19.0) was used.
Sample size calculations were based on the most related study  and our own preliminary supermarket study to achieve power = 0.80, two-tailed α = 0.05, with d as the effect size, to obtain the total sample size n needed, using G*Power (3.1.2). For purchasing, the study by French et al.  employed a 50% discount of fruit and vegetables in a school cafeteria, which led to an increase in purchases of fruit, d = 3.47, and vegetables, d = 0.86, relative to baseline. Based on these results, a sample size of n = 6 would be needed for fruit purchases and n = 46 would be needed for vegetable purchases. On the basis of our preliminary study with n = 18, d = 1.37, for purchasing of discounted vs. nondiscounted foods, a sample size of n = 16 would be needed. For fruit and vegetable intake in that preliminary study, d = 0.79, and n = 42 would be needed. For changes in body weight in that preliminary study during the intervention between the discount and control groups, d = 0.79, and n = 42 would be needed. Thus, our sample size n = 47 was expected to be adequate to reveal significant differences in purchasing, food intake, and body weight.
Purchasing of fruits and vegetables
The repeated measures ANOVA over the three time periods of baseline, intervention, and follow-up, showed a significant effect of time on gross weekly purchasing of discounted fruits and vegetables, F(2,90) = 7.3, P = 0.001, d = 0.81, and a significant group-by-time interaction, F(2,90) = 15, P < 0.0001, d = 1.2 (Figure 3). During the intervention period, gross weekly purchasing of discounted fruits and vegetables was more than three times greater by the discount group ($5.50, SD = $3.22) than the control group ($1.69, SD = $2.07). The average cost of discounts during the intervention period per participant in the discount group was $2.75 per week, which would amount to $143 a year.
Post hoc tests showed that the discount group increased gross spending on discounted fruits and vegetables from the baseline period to the intervention period (t = 6.3, P < 0.0001, d = 1.24), and decreased from the intervention to the follow-up (t = −5.6, P < 0.0001, d = 0.88) to a value not significantly different from the baseline period (P = 0.26). For the control group, however, gross spending on discounted fruits and vegetables did not change significantly from the baseline period to the intervention period (P = 0.12) or from the intervention period to the follow-up period (P = 0.31). Post hoc comparisons showed a significant group-by-time interaction, between the baseline and the intervention periods, F(1,45) = 26, P < 0.0001, d = 1.5, between the intervention and follow-up, F(1,45) = 11, P = 0.002, d = 0.98, and between the baseline and follow-up, F(1,45) = 7.2, P = 0.01, d = 0.80, indicating a partially sustained effect of the discount on purchasing during the follow up period.
Purchasing of noncaloric beverages
Over the three time periods, there was no significant effect of time on gross weekly purchasing of discounted noncaloric beverages, F(2,90) = 1.0, P = 0.37, d = 0.30, and no significant group-by-time interaction, F(2,90) = 2.2, P = 0.11, d = 0.45.
Fruit and vegetable intake from 24-hr recalls
Over the three time periods, there was no significant effect of time on fruit and vegetable intake, F(2,76) = 1.7, P = 0.18, d = 0.43, but the group-by-time interaction approached significance, F(2,76) = 2.8, P = 0.067, d = 0.54 (Figure 4). Post hoc tests showed a significant group-by-time interaction between the baseline and intervention periods, F(1,45) = 4.6, P = 0.037, d = 0.64, but not between the intervention and the follow-up, F(1,38) = 0.39, P = 0.85, d = 0.06, and also a significant interaction between the baseline and follow-up, F(1,38) = 3.9, P = 0.05, d = 0.64, indicating a sustained effect of the discount on intake during the follow-up period. The fruit and vegetable intake in the discount group increased from 135 g (1.7 servings), SD = 173 g, during the baseline period to 252 g (3.2 servings), SD = 250 g, during the follow-up period, while the intake in the control group did not differ significantly between 154 g (1.9 servings), SD = 147 g, during the baseline period and 145 g (1.8 servings), SD = 162 g, during the follow-up period (Figure 4).
The intake of discounted fruits and vegetables within the discount group during the intervention period correlated significantly with gross weekly purchasing of the discounted items (r = 0.62, P = 0.0004) during that period. By contrast, the intake of discounted fruits and vegetables within the control group during the intervention period was not correlated with gross weekly purchasing of the discounted items (r = −0.14, P = 0.56).
Noncaloric beverage intake from 24-hr recalls
Over the three time periods, there was no significant effect of time on noncaloric beverage intake, F(2,76) = 0.034, P = 0.97, d = 0.059, and no significant group-by-time interaction, F(2,76) = 2.3, P = 0.10, d = 0.50.
Energy intake from caloric beverages
There was no significant effect of time on energy intake from caloric beverages, F(2,76) = 0.097, P = 0.91, d = 0.10, and no significant group-by-time interaction, F(2,76) = 1.3, P = 0.28, d = 0.37.
Total energy intake
There was no significant effect of time on total energy intake, F(2,76) = 0.92, P = 0.41, d = 0.31, and no significant group-by-time interaction, F(2,76) = 0.14, P = 0.87, d = 0.12.
Over the five time points of measurements, there was a significant effect of time on body weight, F(4,136) = 3.5, P = 0.009, d = 0.64, but no significant group-by-time interaction, F(4,136) = 0.97, P = 0.43, d = 0.34 (Figure 5). Within-group post hoc tests revealed a significant decrease in body weight in the discount group from the beginning to the end of the intervention period (−1.1 kg, SD = 1.8 kg), t = −3.0, P = 0.006, d = 0.44, and from the beginning of the intervention period to the end of the follow-up period (−1.3 kg, SD = 2.3 kg), t = −2.6; P = 0.016, d = 0.40. Within the control group, the changes in body weight were not significant from the beginning to the end of the intervention period (P = 0.69) and from the beginning of the intervention period to the end of the follow-up period (P = 0.26).
Over the five time points, there was a significant effect of time on body fat (kg), F(4,108) = 2.5, P = 0.044, d = 0.61, but there was no significant group-by-time interaction, F(4,108) = 1.1, P = 0.35, d = 0.41. Within-group post hoc tests revealed no significant changes in body fat in the discount group from the beginning to the end of the intervention period (P = 0.17) or from the beginning of the intervention to the end of follow-up (P = 0.12). The changes in body fat within the control group were also not significant from the beginning to the end of the intervention period (P = 0.69) or from the beginning of the intervention to the end of the follow-up (P = 0.17). Over these time points, there was also no significant effect of time on fat-free mass (kg), F(4,108) = 1.4, P = 0.24, d = 0.45, and no significant group-by-time interaction, F(4,108) = 0.40, P = 0.81, d = 0.24.
Dropouts did not differ from completers in body weight at the beginning of the baseline period (P = 0.11), or in fruit and vegetable intake (P = 0.086) and sugar sweetened beverage intake (P = 0.64) during the baseline period. Dropouts, however, did have a significantly higher total energy intake during the baseline period (2,895 kcal, SD = 1,036 kcal) than completers (2,218 kcal, SD = 918 kcal), F(1,54) = 0.40, P = 0.05, d = 0.54.
The results showed that the discount group purchased more fruits and vegetables during the 8-week intervention period than did the control group. During the follow-up period when discounts ceased, the discount group reverted towards baseline purchasing, confirming that the intervention was responsible for the change in purchasing. Purchasing, however, still remained significantly above the baseline period, reflecting a partially sustained effect of the intervention.
Based on 24-hr recall data of fruit and vegetable intake, the discount group also increased fruits and vegetable intake, as compared to the control group, from the baseline period to the intervention period, which was then maintained during the follow-up period, indicating a sustained effect. The significant correlation within the discount group between gross weekly purchasing of the discounted items and the intake of fruits and vegetables during the intervention period is evidence that increased purchasing led to increased fruit and vegetable intake.
The 24-h recall data of beverage intake, however, did not show group differences in noncaloric beverage intake or in energy intake from caloric beverages, consistent with a lack of group differences in the purchasing of noncaloric beverages. This lack of group differences in intake could be due to participants not purchasing much bottled water or diet sodas for consumption, and maintaining their drinking of tap water or regular caloric sodas, even with noncaloric beverages being discounted. The lack of differences in intake could also be due in part to the participants not being able to report accurately the frequency and volume of water consumed and/or due to their failure to distinguish between tap water and the discounted supermarket bottled water.
The total energy intake based on 24-h recall data did not differ significantly over time or between groups over time. The 24-h recalls, however, each represent intake from 1 day only, and it is possible that small daily differences in intake could accumulate over time and lead to weight change.
Body weight decreased over time across groups, which may have been due in part to participants being in a study that tracked their dietary intake and body weight. Given that the weight change did not differ significantly between groups, the study falls short of demonstrating a weight loss effect, even though there was significant weight loss within the discount group. It may take longer than 8 weeks for changes in purchasing behavior to alter total energy intake, which would translate into changes in body weight between groups.
Body fat also decreased over time across groups, but unlike body weight, the decrease in body fat was not significant within the discount group. This may be due to the limitations of BIA for the measurement of body fat.
The findings from this study extend results from other studies. Implementation of a 50% price reduction of fruit and salad in a workplace cafeteria noted a threefold increase in sales . Likewise, a 50% price reduction of fruits in a school cafeteria quadrupled sales . Lastly, a 50% price reduction of low-fat (≤3 g fat) snacks in vending machines increased sales nearly twofold [17, 18].
In general, the long term economic benefits of lower body weight and improved health has been evaluated to greatly exceed the medical costs and the costs of reduced work productivity associated with obesity . According to the World Health Organization (WHO), 1.7 million (2.8%) of deaths worldwide can be attributed to low fruit and vegetable intake . With the current intervention, we observed that discounting fruits and vegetables by 50%, cost on average only $2.75 per week per person, which translates to an intake of one extra serving of fruits and vegetables per day per person.
The results from this study are supportive of recommendations to implement price reduction of fruits and vegetables. More long-term research may be needed to provide adequate evidence to justify government initiatives to provide fruit and vegetable subsidies.
Limitations of the study include a moderate sample size (n = 47) from a specific neighborhood (Upper West Side of Manhattan), which may not generalize fully to other populations. To help ensure a clear relationship between participants' food purchasing and intake, only small household sizes were included, also limiting generalizability. Participants were assumed to have complied with the contracted study requirements, including an agreement to shop exclusively at the designated supermarket, but compliance was not verified. It is also possible that even with perishable fruits and vegetables, some stockpiling of discounted fruits and vegetables may have occurred during the intervention period, which carried over to the follow-up period and contributed to sustainment in fruit and vegetable intake. Even though body weight did not decrease significantly in the discount group relative to the control group, it is possible that significant decreases would be observed with a longer intervention period. Additionally, the increased fruit and vegetable intake could have led to reductions in other markers, such as fasting glucose, insulin, LDL and HDL cholesterol, and triglycerides, which were not measured in this study.
In summary, the results showed that a supermarket-based 50% discount of low-ED fruits and vegetables, led to more than threefold purchasing by the discount group relative to the control group. This increased purchasing translated into a 50% increase in fruit and vegetable intake by the discount group, which was maintained during the follow-up period. These positive findings lend support to the developing environmental approaches to help reduce the prevalence of overweight and obesity, and to the testing of further initiatives by private groups or the government to subsidize low-ED fruits and vegetables either at the supermarket level or earlier in the process at the agricultural level.
We thank Nicholas D'Agostino III, current Chairman and COO of D'Agostino Supermarkets Inc., who provided access to market resources (basement office space, file cabinet, etc.) and graciously declined reimbursement for the incurred cost of discounting food items. We also thank Anderson Chung and Scott Allen for collecting, storing, and analyzing purchasing data. Research assistants Beatriz Cole, John Zaravinos, Shia Bochner, Sean Teagarden, and Charlie Baley recruited and screened participants, performed assessments, and served as liaisons between St. Luke's Hospital and D'Agostino Supermarkets. The authors report no conflicts of interest.
- 1Obesity and overweight: What are overweight and obesity. Fact sheet N°311. Geneva: World Health Organization; 2006..
- 5Behavioral treatment of obesity: New approaches to an old disorder. In: Goldstein D (ed.). The Management of Eating Disorders. Totowa, NJ: Humana Press; 1999:173-199., .
- 9The environment and obesity. In: Fairburn CG, Brownell KD (ed.). Eating Disorders and Obesity: A Comprehensive Handbook. New York, NY: Guilford Press; 2002. p 433-438..
- 13Weighing the Options: Criteria for Evaluating Weight Management Programs. Washington, DC: National Academy Press; 1995..
- 14Three-year follow-up of Pawtucket Heart Health's community-based weight loss programs. Am J Health Promot 1993;7:182-187., , , et al.
- 19Three-year follow-up of Pawtucket Heart Health's community-based weight loss programs. Am J Health Promot 1993;7:182-187., , , et al.
- 24Dietary energy density and weight regulation. Nutr Rev 2001;59:192-199..
- 32Bureau USC. Census 2000 Summary File 3, 2000. Accessed Sep 8, 2011, from http://www.census.gov/census2000/sumfile3.html.
- 33Measuring intake of fruit and vegetables. Background paper for the Joint FAO/WHO Workshop on Fruit and Vegetables for Health, 1-3 September 2004, Kobe, Japan..
- 34Effectiveness of interventions and programmes promoting fruit and vegetable intake. Background paper for the Joint FAO/WHO Workshop on Fruit and Vegetables for Health, 1–3 September 2004, Kobe, Japan., , , et al.
- 35Statistical Power Analysis for the Behavioral Sciences. New York: Academic Press; 1977..
- 37WHO. Global strategy on diet, physical activity and health. Promoting fruit and vegetable consumption around the World. 2006. www.who.int/dietphysicalactivity/fruit/en/index.html, accessed Feb 2013.