A Population-Based Nutrition Intervention in College Students

Coronary heart disease (CHD) is a leading cause of death in young adults and at least half of college students ages 18-24 have CHD risk factor. Unhealthy dietary choices made by college students contribute to the development of CHD risk factors. Eighty-percent of heart disease is preventable through diet and lifestyle and college students are ideal targets for prevention efforts since they are in the process of establishing lifestyle habits, which track forward into adulthood. The purpose of this dissertation is to provide evidence for the need to target this age group before disease progression occurs and to present the results of a population-based intervention to increase whole grains and improve CHD factors in college students. Manuscript 1 “Coronary Heart Disease Risk Factors in College Students” is a narrative review paper highlighting the need for improved heart disease risk assessment and awareness in college students. This review provides pathological evidence along with current risk factor prevalence data to demonstrate the need for early detection. The impact of diet is addressed and population-based strategies are presented as cost-effective ways to produce wide-scale risk reduction. Manuscript 2 “A Population-Based Nutrition Intervention to Increase Whole Grain Intake in College Students” is a primary research paper on the impact of a nutrition messaging intervention in campus dining halls. Results indicate that a 6-week messaging intervention in campus dining halls had a positive impact on whole grain consumption and on HDL-C in college students. Future research should focus on population-based approaches on college campuses to prompt students to make healthier selections.


LIST OF TABLES
Coronary heart disease (CHD) risk in young adults, ages 18-24, is underestimated despite the high prevalence of CHD risk factors (1-4) and early signs of atherosclerosis in this age group (5,6). Obesity has more than doubled in children and more than tripled in adolescents over the past 30 years (7). This weight gain tracks forward and worsens in young adulthood (8). Heart disease risk increases by 2-4% for each year a young adult is obese (9). As many as 33% of young adults are overweight (1) and this excess weight leads to dyslipidemia (10) and increases in metabolic syndrome (11), diabetes (12) and CHD (3) risk. Coronary heart disease accounts for 50% of cardiovascular disease (CVD) deaths and is one of the leading causes of death in young adults (13). Coronary heart disease costs the US $108.9 billion each year in health care services, medications and lost productivity (14), which is more than any other disease. A death occurs from CVD every 40 seconds in the US, which would wipe out a college campus of 25,000 in less than 12 days (15).
More than half of young adults have at least one CHD risk factor and this greatly increases lifetime heart disease risk (16). Since many CHD risk factors surface in adolescence (13,(17)(18)(19) and track forward to adulthood (20), the American Heart Association's (AHA) 2020 Strategic Impact Goals along with the National Heart, Lung and Blood Institute's (NHLBI) 2012 Expert Panel on Integrated Guidelines for Cardiovascular Health and Risk Reduction in Children and Adolescents (21) emphasize primordial prevention beginning in childhood and adolescence (16). This concept of primordial prevention was introduced by Strasser in 1978 (22) and focuses on preventing the development of risk factors themselves (16). Dietary modifications are central to this approach (16).
Despite screening recommendations for all adults over age 20 (23,24), < 50% of women and < 40% of men of this age are screened for CHD risk (25). In addition, the majority of young adults are unaware of their risk (26). Until primordial prevention strategies are implemented to avoid risk factor development in the first place, there is a need for improved screening, risk assessment, management and education in this age group. Early detection and intervention are critical since 80% of CVD events are preventable through diet and lifestyle (27). Diets low in saturated fat and high in fruits and vegetables reduce the risk of new cardiac events by 73% (28).
Despite this evidence, young adults have high intakes of solid fats, added sugars (29) and sodium (1, 30), along with inadequate intakes of fruits and vegetables (31), whole grains (32, 33) and fiber (30). The AHA recently issued a scientific statement recommending reductions in added sugar intake in response to research linking sugar to excess energy intake, obesity, dyslipidemia and CHD risk (34). Sugar consumption has increased by nearly 20% from 1970 to 2005, supplying almost 500 kcal/day (35).
Adolescents consume more sugar than any other age group (549 kcals) (34) and this continues into young adulthood (29). Collectively, these poor dietary choices contribute to the high prevalence of CHD risk factors in this age group (36-39).
In The purpose of this review is to demonstrate the need for improved screening and risk awareness of CHD in young adults by revealing pathological changes that start in childhood and manifest themselves in young adult CHD risk factors. In addition, successful population-based prevention/treatment strategies used in other populations will be discussed with a focus on how these strategies can be applied to this age group.

Childhood Risk Factors Correlated with Extent of Lesions
Research indicates that atherosclerosis has childhood roots. In the 1950s and This autopsy study showed that the extent of atherosclerotic lesions was directly correlated to antemortem levels of TC, TG, LDL cholesterol (LDL-C), HDL cholesterol (HDL-C), BP, BMI and cigarette smoking in young adults (6,49 As many as 10-20% of young adults have advanced atherosclerotic lesions (53). This progression is correlated with the number of CHD risk factors; young adults with ≥ 3 childhood risk factors had a 9-fold increase in atherosclerotic plaque area compared to those with none (6). As shown in Table 1, risk factors in childhood were shown to be strong predictors of preclinical atherosclerosis even after adjustment for adult risk factors (54, 55). These findings are critical from a prevention standpoint as those at risk of developing atherosclerosis can be identified and treated decades before clinical manifestation of disease.

Childhood Risk Factors Associated with Preclinical Disease Markers
Hyperlipidemia early in life is directly related to pathologic changes and functional abnormalities and strongly predicts CHD in adulthood (56 "all-you-can-eat" pricing. This "all-you-can-eat" environment and the wide variety of foods available in dining halls leads to larger portion sizes, increased energy intake and weight gain (79). In the first semester, college students gain weight up to 11 times faster compared to young adults not in college (71) and maintain this weight throughout college (80) and into adulthood. This additional weight, most of which is excess body fat, can lead to dyslipidemia and increased heart disease risk (10).

Prevalence of CHD Risk Factors in College Students
Coronary heart disease risk factors in young adulthood can be the result of pathological changes from childhood. Only 20% of CHD in young adults is related to non-atherosclerotic factors (41). Results from the few cross-sectional studies that have assessed CHD risk in college students, ages 18-24 years show an alarmingly high prevalence of young adults with abnormal risk factor profiles ( Table 2). Huang et al.
(81) reported that the most prevalent risk factors in a sample of 163 college students were elevated TC (12%) and low HDL-C (14%). Impaired glucose metabolism was also a concern as just over 6% had pre-diabetes. Overweight students had worse risk factor profiles (waist circumference (WC), BP, TC, LDL-C, VLDL cholesterol (VLDL-C), TG, leptin, insulin) compared to normal weight students and were nearly 3 times more likely to have at least one metabolic syndrome component. 189 first year college students and found that 18% had elevated TG and 20% had low HDL-C for gender. Metabolic syndrome risk was also high; 28% met at least one of the criteria for metabolic syndrome and 4% had metabolic syndrome. Obese students were more likely to meet 3 or more metabolic syndrome criteria and had a higher prevalence of abnormal HDL-C, WC and BP compared to subjects with a BMI<30 kg/m 2 . Gender differences were also noted, with males having a higher prevalence of risk factors ( Table 2).
In a similar study by Huang et al. (4) that examined prevalence of metabolic risk and gender differences in a sample of 300 students, 24% had low HDL-C, 9% had elevated fasting glucose and 9% had elevated TG. Overall prevalence of metabolic syndrome was low (1%) but 1/3 of the sample had at least one component. As shown in Table 2, males had a worse metabolic profile than females.
In a larger study performed on 1,701 college students, Burke et al.
Males also had a worse risk factor profile (BMI, glucose, TC, HDL-C, LDL-C, SBP and diastolic blood pressure (DBP)) than females in this study. In a subsequent analysis of the same data but with a larger sample size, (n=2,103) nearly 1/3 had low HDL-C, nearly 2/3 had high BP and approximately 1/4 had elevated TC or LDL-C (3).
Metabolic syndrome was observed in up to 10% of the sample and those with a higher BMI had a significantly greater number of individual metabolic syndrome risk factors.
The differences in prevalence rates across studies can be partially attributed to demographic differences between universities. Risk factor profiles can be expected to vary due to different ethnic breakdowns and lifestyle factors across geographically dispersed university samples (2). There were also gender differences; a higher prevalence of CHD risk factors was found in men. Risk factor profiles were worse in overweight and obese individuals, regardless of gender. Collectively, these studies demonstrate that dyslipidemia and metabolic dysfunction are a common and major concern in young adults. As previously discussed, poor dietary choices made by this age group contribute to the high prevalence of risk factors. These data underscore the need to identify those at risk, especially male and overweight/obese young adults, so that steps can be taken to prevent future CHD risk and manage existing risk factors.
Data collected to-date demonstrates that college students are at risk for heart disease but additional research needs to be done on young adults not in college to get a more comprehensive profile of this age group.

Historically Conflicting Guidelines
Data from the cross-sectional studies mentioned above demonstrate that CHD risk factor prevalence is high in this age bracket, yet universal risk assessment for primordial and primary prevention is lacking. Although the importance and need for screening for early detection and management of dyslipidemia is recognized from public health organizations, including the NHLBI, AHA, American Academy of Pediatrics (AAP), and US Preventive Services Task Force (USPSTF), the majority of young adults are not screened (25). The absence of apparent disease in young adults contributes to the underestimation of risk in this age group by both young adults themselves and health professionals (26,82,83). This underestimation of risk and historically differing risk assessment guidelines contribute to this problem (84). More recent 2013 ACC/AHA CVD Assessment Guidelines also support the need for risk assessment early in life to motivate lifestyle changes in younger individuals who may be at low short-term risk but could benefit from long-term risk assessment. Long-term risk assessment of traditional CVD risk factors is recommended every 4-6 years beginning at age 20 for those who are free from atherosclerotic cardiovascular disease (23).

Inadequate Screening in Young Adults
National Health and Nutrition Examination Survey data from 1999-2006 on 2587 young adults ages 20-45 years, indicated that 2/3 have at least one CVD risk factor. This is alarming since less than 50% of females and less than 40% of males reported being screened prior to the assessment visit. The screening rate for young adults in the 18-24 year age bracket can be expected to be even lower as screening rates increase with age (99). Younger males, in particular, are more than 50% less likely than their female counterparts to obtain preventive services (100). Data from NHANES show that women are more likely to have health insurance and see a healthcare provider (25). These low screening rates are especially concerning among young adults with multiple risk factors as the extent of atherosclerosis is directly correlated with the number of risk factors.
The AHA supports population-based strategies such as screenings at universities to identify at risk individuals (16,97,101). Policy changes are needed to promote increased screening in primary care settings, clinics, schools, worksites and community sites. These screenings are particularly important in the young adult age group that may go otherwise undetected by the health care system (102)  research is needed to better understand and eliminate barriers to screening. This needs to be done at the policy, provider and patient level to improve suboptimal screening in young adults (104).

Population-Based Nutrition Interventions in College Students
Until primordial prevention strategies are successful in avoiding risk factor development all together, risk factor screening needs to work in tandem with education and management for effective disease prevention. Strategies that focus on mg/dL in 1987 to 182 mg/dL in 1992, decreasing the prevalence of hypercholesterolemia from 25% to 6% in men and from 22% to 5% in women (106,107). This intervention was a classic example of a population-based strategy that effectively shifted the entire distribution of risk. Estimates from the World Heart Federation show that a universal reduction in sodium intake by 1 gram/day would lead to a 50% reduction in the number of individuals needing treatment for hypertension, a 22% decrease in deaths from stroke and a 16% drop in deaths from CHD (28).
Similar population-based strategies can be applied to the college setting.
Although cafeterias can contribute to an obesogenic environment on college campuses, they also represent an opportunity to influence students' diets for the better by providing nutrition information to guide healthy choices (108    The few cross-sectional studies that have assessed CHD risk in college students ages 18-24 demonstrate the need for increased screening, risk assessment and disease prevention in this age group [11][12][13][14] . More than 50% of college students have elevated low-density lipoprotein cholesterol (LDL-C) 15 . Additionally, as many as Typical diets consumed by college students, which are high in saturated fat 19 and low in whole grains 20 negatively affect these risk factors, especially LDL-C.
More than 70% of college students exceed total and saturated fat recommendations 19 .
They also fail to meet whole grain recommendations 20

METHODS:
Design A quasi-experimental design was used to assess the impact of a populationbased whole grain and low-fat dairy intervention on CHD risk factors. Baseline and post-intervention assessments were conducted immediately before and after the six- week intervention and the follow-up assessment occurred six months after the baseline assessment. A subsample (n=26) was recruited for Heart Start II, which involved additional measures. All measurements described were obtained at baseline, postintervention and follow-up.

Sample
Participants were recruited via classroom announcements at a medium sized northeastern university. Eligible participants were 18-24 year old males and females with a campus meal plan and a BMI ≥18.5 kg/m 2 . Exclusion criteria included being pregnant or lactating, or self-report of one of the following conditions: eating disorder, liver disease, bleeding disorder, diabetes, cancer, or CHD. All participants read and signed an informed consent approved by the University's Institutional Review Board.

Measures Dietary Intake
The from the mean values from the three 24-hour recalls for Heart Start II participants to assess diet quality in Heart Start II participants (n=26).

Biochemical
Following a 12-hour fast, finger sticks were performed on all participants to obtain blood samples for determination of blood lipid and glucose concentrations.
Values for LDL-C, TC, TAG, HDL-C and glucose were obtained using Cholestech LDX table-top analyzers (Cholestech, Hayward, CA).

Anthropometrics
Height was measured to the nearest 0.1 cm using a Seca 220 stadiometer (Seca Corporation, Hamburg, Germany). Weight was measured to the nearest 0.1 kg using a calibrated digital Seca 769 scale (Seca Corporation, Hamburg, Germany).
Measurements were taken in duplicate and the average of the two values was used for the analysis. Body mass index was calculated using the following formula: weight in kilograms/height in meters 2 . Waist circumference was measured in duplicate at the top of the iliac crest upon exhalation to the nearest 0.1 cm using a Gulick fiberglass, non-stretchable tape measure with an attached tensometer (Patterson Medical, Mount Joy, PA). The average of the two values was used for the analysis.

Blood Pressure
Blood pressure was measured after a 5 minute seated rest period using an automatic blood pressure monitor with arm cuff (Omron HEI-711, Omron Health Care Products, Issaquah, WA). Measurements were re-taken two minutes apart until values were within 2 mmHg. The average of the two values in agreement was used for the analysis.

Intervention
Heart Start I and II participants were exposed to a 6-week intervention which consisted of benefit-based nutrition messages in the two main campus dining halls (Hope and Butterfield). Messages were displayed on television monitors and on pointof-selection signs at the deli and dairy stations in both dining halls. Prompts to choose whole grain bread were also verbally provided by the deli station staff in both dining halls. Additionally, nutrition education booths to promote whole grain and low-fat dairy consumption were positioned in a high traffic area outside of Hope. Message and booth content alternated between whole grains and low-fat dairy each week. Students with meal plans were able to eat at either dining hall and all students who ate at the dining halls were exposed to the intervention.
Intervention materials addressed specific motivators of healthy eating (increased energy, healthy body weight and staying full) from previously conducted focus groups 43 . Additionally, Heart Start II participants received the same nutrition message that was displayed on the television monitors in the dining halls each weekday via text message or email, depending on their preference. Google Voice (Google, Mountain View, CA), a web-based application, was used to deliver text messages.

Analysis
Descriptive statistics were performed and skewness and kurtosis were examined to determine data distribution. Non-normally distributed data were transformed. Body mass index, LDL-C, total grains, low-fat dairy and soluble fiber were log transformed. Triacylglycerides and sugar-sweetened beverages (SSB) were square root transformed. Whole grains, semi-whole grains, total fiber, reduced fat dairy, glucose and systolic blood pressure (SBP) were analyzed using non-parametric tests. Continuous variables were expressed as mean ± standard deviation and categorical variables were expressed as frequencies. Repeated measures analysis of variance with post hoc tests using the Bonferroni adjustment were used to determine if there were significant differences over time. The Friedman test with post hoc Wilcoxon signed rank tests using a Bonferroni adjusted alpha value were used to assess differences over time for whole grains, semi-whole grains, fiber and glucose.
Mixed between-within analysis of variance assessed differences between groups over time. Chi-square tests were used to analyze categorical variables. Statistical significance was set at p<0.05 for all tests.

RESULTS:
Participant characteristics at baseline are presented in Table 1. The majority of the sample was female (78%) and Caucasian (81%). The mean age was 18.2 ± 0.6 years. At baseline, more than 50% of females and 36% of males had low HDL-C for gender (<40 M, <50 F mg/dL), 19% had elevated LDL-C (≥100 mg/dL), 14% had elevated TAG (≥150 mg/dL) and 13% had elevated SBP (≥130 mmHg). More than 80% of the sample had never or were unsure as to whether they ever had their cholesterol checked. Sixty-three Heart Start I participants completed all three assessment visits and 18 of these 63 completed additional measurements for Heart Start II.
Data from the NCI Screener indicated that whole grain intake increased over time (χ 2 (2, n=69) = 10.6, p=0.005). Whole grain intake increased from baseline to follow-up (0.8 ± 1.1 oz to 1.1 ± 1.5 oz, p=0.008) and from post-intervention to followup (0.8 ± 0.8 oz to 1.1 ± 1.5 oz, p=0.006). Purchasing record data (used as a proxy for consumption) indicated that percent whole grain consumption doubled (12.7% to 23.9%) in the dining hall with nutrition education booths, point-of-selection signs, promotion by deli counter staff and messaging on television monitors (Hope) during the 6 week intervention (data not shown). In Hope, baseline whole grain consumption was significantly lower than the intervention and follow-up period but not different from post-intervention. In Butterfield, whole grain consumption significantly increased across baseline, post-intervention and follow-up periods and was higher than consumption at Hope at all time points.
As displayed in Table 2 There was no primary measure of low-fat dairy intake. However, purchasing records used as a proxy for dairy intake indicated that nonfat dairy increased by 3-4% during the intervention and were significantly higher at follow-up compared to other time points in Hope (data not shown). Data from the NCI screener showed that total dairy intake decreased over time (Wilks' Lambda = 0.85, F 2, 69 = 6.16, p=0.003, η 2 =0.15). Sugar-sweetened beverages (SSB) also significantly decreased over time Twenty-four hour recall data showed no changes in total grain, semi-whole grain, refined grain, fiber, total dairy, full fat dairy, reduced fat dairy, low-fat dairy or saturated fat intake in Heart Start II participants (p>0.05 A post-intervention survey revealed that nearly 80% of participants noticed the messages. Seventy percent reported that the messages prompted them to choose whole grains, while only 40% indicated that the messages prompted them to choose low-fat dairy. Point-of-selection messaging was the most effective messaging delivery method for both whole grains and low-fat dairy (Figure 1).

DISCUSSION:
The results of this study demonstrate that population-based POS messaging in campus dining halls is an effective strategy to increase whole grain intake in college students. Improvements in HDL-C were seen. Declines in total dairy intake over time suggest that the focus of interventions should shift from low-fat dairy to total dairy.
Whole grain consumption (as measured by the NCI screener) increased by nearly 40% from baseline to follow-up. This is supported by the purchasing records, which indicated that percent whole grain consumption doubled during the 6-week intervention. It is also consistent with the findings from pilot testing in the spring of 2012 that showed a 12% increase in whole grains when messages were displayed in dining halls for one-week (S. Mello, personal communication). Results from this study suggest that sustained messaging is needed to produce lasting behavior change as whole grain consumption returned to baseline levels after the messages were removed.
Although increases in whole grain consumption were observed, the mean intake at follow-up still fails to meet recommendations. This is consistent with findings by Ha et al. 20 that reported an increase in whole grain consumption in college students after a whole grain intervention embedded in a semester-long nutrition course significantly increased whole grain intake from 0.37 oz to 1.16 oz. Despite this increase, whole grain intake after the intervention was >50% less than the minimum recommendation of 3 oz. Exploratory analyses on Heart Start II participants who received the additional text messages indicated that this subgroup had non-significant increases in their whole grain intake. This may be attributed to the small sample size in this subgroup analysis (n=18).
Baseline CHD risk factor prevalence data was similar to previous crosssectional estimates of CHD risk factors in this age group [12][13][14]44 . Weight gain during the first year of college is well documented 46-51 . Our sample gained less weight than has been previously reported in this age group 49, 50 .
Although weight status was not a primary aim of the intervention, the weight gain observed in this population highlights the need for weight gain prevention efforts in this age group. However, interventions focusing on weight must be sensitive to the higher prevalence of disordered eating in this age group 52, 53 .
Purchasing records showed a slight increase in non-fat dairy over time but the NCI Screener indicated that total dairy intake decreased over time. Since this screener did not allow for the analysis of components of total dairy (reduced fat, low-fat, nonfat) it cannot be determined whether there was a shift to low-fat dairy over time. A decrease in total dairy, however, is consistent with previous findings 54, 55 and provides evidence for the need for additional efforts in this age group to prevent further declines in dairy intake. A reduction in dairy intake typically coincides with an increase in SSB as a result of displacement 56 . In this sample, however, SSB consumption significantly decreased along with dairy consumption over time.
Decreased consumption of dairy at follow-up may be a function of weight conscious eating behaviors that occur pre-spring break in anticipation of beaches, as dairy is perceived to be "fattening" 32, 57 . Similarly, purchasing records at follow-up showed an increase in whole grain consumption, which may be a function of pre-spring break healthier eating.
Feedback on the individual intervention components revealed that POS messaging was the preferred method of messaging. Point-of-selection messaging has previously been shown to be an effective population-based strategy to promote healthy choices in college dining halls 58 . In a dining hall intervention that utilized signs, Overall, findings from this study indicate that a population-based nutrition intervention was effective in increasing whole grain intake in college students. Future research should focus on implementing population-based approaches to promote healthy eating on college campuses as cost-effective ways to guide students in making better dietary choices.

SO WHAT? Implications for Health Promotion Practitioners and Researchers
What is already known on this topic?

Study Flow Chart:
Data Collection:

Online Surveys
Students who were interested in participating emailed the study email address and were then sent a link to complete the online surveys before the initial assessment visit. Green eating includes participating in most of the following behaviors: • Eating locally grown foods, produce that is in season and a limited amount of processed food • Consuming foods and beverages that are labeled fair trade certified or certified organic • Consuming meatless meals weekly and (if consuming animal products) selecting meats, poultry and dairy that do not contain hormones or antibiotics.
Based on the definition of green eating, which of the following best describes you now: • I do not regularly practice green eating and do not intend to start within the next 6 months (precontemplation) • I am thinking about practicing green eating within the next 6 months • I am planning on practicing green eating within the next 30 days • I regularly practice green eating and have been doing so for less than 6 months (action) • I regularly practice green eating and have been doing so for 6 months or more

(maintenance)
All participants read and signed an informed consent approved by URI's Institutional Review Board. All measurements were obtained at baseline, post-intervention and follow-up:

Anthropometrics
Height was measured to the nearest 0.1 cm using a Seca 220 stadiometer (Seca Joy, PA). The average of the two values was used for the analysis.

Biochemical
Following a 12-hour fast, finger sticks were performed on all participants to obtain blood samples for determination of blood lipid and glucose concentrations.
Values for LDL-C, TC, TAG, HDL-C and glucose were obtained using Cholestech LDX

Blood Pressure
Blood pressure was measured after a 5 minute seated rest period using an automatic blood pressure monitor with arm cuff (Omron HEI-711, Omron Health Care Products, Issaquah, WA). Measurements were re-taken two minutes apart until values were within 2 mmHg. The average of the two values in agreement was used for the analysis.

Dietary Intake
The NHANES 2009-2010 National Cancer Institute Dietary Screener Questionnaire (DSQ) was used to assess intake of fruits and vegetables, dairy/calcium, whole grains/fiber, added sugars, red meat, and processed meat in Heart Start I participants (7). Variables from the survey monkey download were re-named according to the DSQ codebook for the self-administered paper version. Eight-digit food codes were assigned to cereal responses. The SAS program and associated data files were used to analyze the dietary screener questionnaire data file (8). The following variables were calculated from the syntax: predicted fiber (gm) per day, predicted calcium (mg) per day, predicted added sugars (tsp) per day, predicted ounce equivalents of whole grains per day, predicted cup equivalents of dairy per day, predicted cup equivalents of fruits and vegetables (including legumes) per day, predicted cup equivalents of fruits and vegetables (including legumes) except French fries per day and predicted added sugars (tsp) from sugar-sweetened beverages.
Purchasing records from dining services were used as a proxy for whole grain and low-fat dairy consumption. Purchasing records were obtained for bread and dairy products that offered a whole grain or low-fat dairy alternative (bread, rolls, breadsticks, English muffins, milk and yogurt) to determine if students selected the whole grain or low-fat dairy option. Purchasing records were obtained at baseline, intervention, post-intervention and 6-month follow-up. According to the whole grain definition used by dining services, items were categorized "whole grain" if the first ingredient was a whole grain. Dairy products were categorized as follows: whole (full fat), low-fat (1% or 2%) and nonfat (skim). Average values were calculated for individual items at each time point and were used for the analyses.
Twenty-four hour dietary recalls were collected and analyzed for Heart Start II participants (n=26) using the multiple pass method in conjunction with the Nutrition Data System for Research (NDS-R) software (University of Minnesota, Minneapolis, MN) version 2012. All participants completed three 24-hour dietary recalls: one inperson and two over the phone on three non-consecutive days (including two weekdays and one weekend day) (9, 10). Nasco food models (eNasco, Fort Atkinson, WI) and food amounts booklets were available during the initial in-person 24-hour recall to more accurately estimate portion size (11). Participants were given the booklets after the initial recall for the phone recalls. The mean values of the three recalls provided dietary data for analysis.

Intervention
Heart Start I and II participants were exposed to a 6-week intervention, which consisted of benefit-based nutrition messages in campus dining halls. Messages were displayed on television monitors and on point-of-selection signs at the deli and dairy stations in both dining halls. Prompts to choose whole grain bread were also verbally provided by the deli station staff in both dining halls. Additionally, nutrition education booths to promote whole grain and low-fat dairy consumption were positioned in a high traffic area outside of Hope. Message and booth content alternated between whole grains and low-fat dairy each week. All URI students who ate at the dining halls were exposed to the intervention.
Intervention materials addressed specific motivators of healthy eating in for students (increased energy, healthy body weight and staying full) from previously conducted focus groups (15). Additionally, Heart Start II participants received the same nutrition message that was displayed on the television monitors in the dining halls each weekday via text message or email, depending on their preference. Google Voice (Google, Mountain View, CA), a web-based application, was used to deliver text messages.

Analysis
Sample Size G*Power version 3.1.2 was used to calculate sample size. Sample size calculations were performed based on expected changes in LDL-C from a similar study with an effect size of 0.61 (16). Required sample size was determined to be 23, with alpha set at 0.05 to achieve statistical power at the 0.80 level.
Descriptive statistics were performed and skewness and kurtosis were examined to determine data distribution. Continuous variables were expressed as mean ± standard deviation and categorical variables were expressed as frequencies.
Predicted fiber (gm) per day, predicted added sugars (tsp) per day, total servings of low-fat dairy, LDL-C and BMI were log transformed. Predicted calcium (mg) per day, predicted cup equivalents of fruits and vegetables (including legumes) except French fries per day and TAG were square root transformed. Predicted ounce equivalents of whole grains per day, total servings of semi-whole grains, total servings of fiber, total servings of insoluble fiber, total servings of reduced fat dairy, % kcals from alcohol, predicted added sugars (tsp) from sugar-sweetened beverages and glucose were analyzed using non-parametric tests.
Repeated measures analysis of variance was used to determine if there were significant differences over time. Mixed between-within analysis of variance assessed differences between groups over time. Physical activity was included as a covariate as appropriate. Statistical significance was set at p<0.05 for all tests.
increasing your dietary knowledge and learning about your health status. You will receive the results from your assessment visits (height, weight, body mass index, waist circumference, blood lipids and glucose).

Confidentiality:
Your participation in this study is confidential. None of the information will identify you by name. All records will be stored in a locked office that is only accessible to study personnel.
In case there is any injury to the subject: