Examining Eating Rate in Women Recruited From Low-Income Sites

........................................................................................ ii ACKNOWLEDGEMENTS..................................................................... iii PREFACE............................................................................................ v TABLE OF CONTENTS......................................................................... vi LIST OF TABLES................................................................................ viii LIST OF FIGURES................................................................................ ix MANUSCRIPT...................................................................................... 1 ABSTRACT................................................................................ 2............................................................................... 2 INTRODUCTION......................................................................... 4 METHODS.................................................................................. 7 RESULTS.................................................................................. 18 DISCUSSION............................................................................. 21 REFERENCES ........................................................................... 35 APPENDICES..................................................................................... 39 A. REVIEW OF LITERATURE..................................................... 39 B. CONSENT FORM........................................................................................ 67 C. STUDY TEST MEAL DAY PROTOCOL....................................... 70 D. SCREENING INTERVIEW....................................................... 73 E. VISUAL ANALOG SCALES FOR APPETITE................................. 75 F. VISUAL ANALOG SCALES FOR PALATABILITY....................... 81 G. MODIFIED WEIGHT RELATED EATING QUESTIONNAIRE......... 82 H. INTERNATIONAL PHYSICAL ACTIVITY QUESTIONNAIRE...... 84


ABSTRACT
Obesity is a major health problem in the United States. Food insecurity is related to obesity, especially in women. Obesity is associated with a fast eating rate (ER) and failure to reduce ER during meals. The purpose of this research is to measure ER in food insecure women in community settings utilizing a novel, mobile measurement system using laptop computers. Low-income women (n=20), ages 18-65 (mean±SD age= 46.5±13.7 years) with a body mass index (BMI) over 18.5 kg/m 2 (mean±SD BMI= 35.7±6.6 kg/m 2 ) were recruited from a food pantry and free clinic in Rhode Island. The United States Department of Agriculture (USDA) Adult Food Security Survey Module was used: participants scored in the secure (n=0), marginal (n=1), low (n=4), or very low (n=15) range. Scores of high and marginal were categorized food secure (n=1), and of low and very low as food insecure (n=19). Women were video recorded eating a test meal (400g, 842 kcal vegetable frittata), which was covertly weighed pre and post consumption. Eating rate (g/min), duration (min), energy intake (kcals), total intake (g), kcals/minute, bite size (g/bite), and quartile eating rate (bites/quartile) were calculated.
iii ACKNOWLEDGEMENTS This was quite a challenging and rewarding journey, but I am so proud of all that has come of it. I also have so many people to thank, people with whom I have leaned on greatly these past two years. To my major professor and mentor, Dr. Kathleen Melanson, without your support and guidance this thesis would not have been possible, thank you.
Additional thanks to Dr. Geoffrey Greene who helped design this study, provided the means for me to work on it, and answered every question I had along the way. To my outside committee member, Dr. Kathleen Gorman, thank you for always providing a fresh set of eyes to my work and helping me strive to truly understand how complex and multidimensional food insecurity is. when and what was due. Thank you for all the support both school and non-school related, for the endless fun, all the #divarants, inspirational cat pictures, and buch I could ever ask for. This program would never have been the same. We did it! iv Most importantly, the biggest thank you to my family. Thank you for understanding and giving me support when I needed it the most. To my parents, Bill and Carrie, you two have provided me everything I could ever hope for. I am so thankful to have grown up in such a loving environment. To my sister, Kate, and brother, Liam, thank you for always being the best big sister and brother I could ever ask for and always looking out for me, even though I like to act like I am the oldest.
v PREFACE This thesis was written in manuscript format in compliment with the University of Rhode Island's Graduate School Thesis Guidelines. This thesis contains one manuscript titled "Examining Eating Rate in Low-Income Women". This manuscript was written in the form suitable for submission for publication in the journal Appetite.
viii LIST OF TABLES   Quartile results found that Q1 eating rate was slower than Q2 in over half of participants (n=7). Additionally Q4 eating rate was faster than Q2 and Q3 (n=6), suggesting an accelerated pattern. In conclusion, the novel methodology proved difficult and complex, limiting the study. Findings need to be confirmed with a larger sample.

INTRODUCTION
Obesity is a major health concern in the United States, and more than two-third of American adults are overweight or obese. 1 Obesity is associated with increased risk of diabetes mellitus, cardiovascular diseases, cancer, stroke, and other health problems. 2,3 Obesity is the result of excess weight gain, from increased energy intake relative to energy expenditure. 4 Various factors, both controllable and uncontrollable, contribute to obesity. 2 Controllable factors include diet quality, food environment, eating rate, and food security. 5 Uncontrollable factors include race-ethnicity, gender, and genetics. 3 An emerging area of research designed to reduce obesity risk involves slowing eating rate during meals. Eating rate can be measured using several methods. The gold standard for measuring eating rate involves using a Universal Eating Monitor (UEM), which is a laboratory instrument that uses a hidden scale to weigh the plate at fixed intervals during meals, enabling calculations of eating rate. Meals can also be video recorded, and eating rate can be back-calculated from reviewing the video recordings.
Video recordings can be used to calculate eating rate quartiles; however, bite size per quartile cannot be measured from video recordings. The final method includes weighing the test food pre and post to calculate grams consumed as well as measuring the start and finish times to calculate meal duration in minutes. Average intake (grams or kilocalories), eating rate (kcals/min, bites/min, or grams/min) and bite size (grams/bite) can be calculated from this method.
Using such methodologies, past research shows that eating slowly leads to decreases in energy intake (total kilocalories). [6][7][8][9] Several eating rate interventions conducted at The University of Rhode Island in Kingston, RI were successful in significantly lowering eating rate from pre to post-intervention. 7,8,[10][11][12] Each intervention included education sessions designed specifically to reduce eating rate, including bite size and bites per minute. 7,8,10 After several successful interventions, the research, was moved to Providence, RI and women of all income and education levels were recruited. 10 The intervention however was not successful despite its similarities to the previous studies. 10 The researchers concluded that the participants, who reported as low-income, might have been food insecure. 10 However, other factors besides food insecurity may have influenced eating rate in this population and contributed to the unsuccessful intervention. There has been no research to date involving establishment of eating rate patterns or conducting eating rate interventions in low-income populations who are food insecure.
Low income is defined as earning less than 200% of the federal poverty guidelines and food insecurity is defined as a reduction in the quality, variety, or desirability with or without indications of disruptive eating patterns and reduced food intake. [13][14][15][16][17] The United States Department of Agriculture (USDA) describes a range for food security that categorizes individuals as either having high food security, moderate food security, low food security, or very low food security. 13,16 Previously the range was food secure, food insecure without hunger, food insecure with moderate hunger, and food insecure with severe hunger. 13 Food insecurity is a growing epidemic throughout America, and Coleman-Jensen et al. concluded that in 2013, 14.3% of U.S. households were food insecure. 17 In addition, research supports that food insecurity is associated with increased obesity, especially among women. 13,[18][19][20][21][22][23][24][25][26] Higher rates of obesity among food insecure women may be related to poor diet quality, increased consumption of low-cost energy dense foods and sugar-sweetened beverages, limited access to healthy and nutritious foods, limited access to supermarkets, increased stress and anxiety related to food cost and money, and lack of supplemental benefits. 5,[27][28][29][30] The purpose of this research is to examine eating rate in food insecure women.
Past research shows some overweight and obese women eat at a faster eating rate and as the meal progresses do not slow down but rather continue at a linear or constant eating rate. [31][32][33] Some normal weight women eat at a slower eating rate compared to overweight and obese women and further slow down in eating rate as the meal progresses and satiation begins. 31 No research to date has involved measuring eating rate in food insecure women. It is hypothesized that food insecure women will exhibit a faster average eating rate (grams per minute) when compared to food secure women. In addition, it is hypothesized that food insecure women's quartile eating rate (bites per quartile) will continue at a linear eating rate until meal completion.

East Bay Food Pantry Protocol
Prior to screening participants, the laptop computer was set up at the corner of the Eating rate was also calculated by total grams/ meal duration (minutes) = grams per minute. Additionally kilocalories per minute was calculated using the equation: energy intake (kilocalories)/ meal duration (minutes). Eating rate was calculated from the pre and post method of weighing the test meal before and after and recording meal duration, not measured directly from the video recordings. The total number of bites throughout the meal duration was examined from reviewing the video recordings and from that average bite size was calculated using the equation: total grams/ total bites = gram per bite.
Quartiles were used to examine the eating rate pattern across the test meal. Quartile eating rate was calculated as bites per quartile and was directly measured from video recordings. Quartile1 was considered the beginning of the meal and determined initial eating rate, quartile 2 and quartile 3 were mid-meal, and quartile 4 was the end of the meal and used to determine when eating ends. Eating rate, bites per quartile, were compared between quartiles 1 and quartiles 4 in order to examine eating rate differences from the start of the meal to the finish.

Food Insecurity
A research assistant scored each USDA Adult Food Security Survey Module using the USDA coding sheet (Appendix J). Each participant's raw score ranged between 0 and 10. Participants' raw scores placed them into one of four ranges: high food security (score of 0), marginal food security (score of 1-2), low food security (score of 3-5), and very low food security (score of 6-10). If participants scored high food security or marginal food security they were considered food secure. If participants scored low food security or very low food security they were considered food insecure.
Data were examined for normality using skewness and kurtosis, Shapiro Wilk test, and by examining the histograms, Q-Q plots, and box plots. Skewness and kurtosis were examined by dividing the statistical value by the standard error and results between +/-1.96 were considered normally distributed. Statistical values above 0.05 were considered normally distributed for the Shapiro Wilks test.
Planned comparisons between food insecure and food secure groups could not be performed because only one participant scored food secure. However, exploratory hypotheses were examined. Differences were established using chi-square analysis for categorical variables such as ethnicity, and independent t-test analysis for continuous variables, such as eating rate, energy intake, total grams, meal duration, kilocalories per minute, and quartile eating rate. Analysis of variance (ANOVA) was used to examine differences between the food security ranges (food secure, marginally food secure, low food secure, and very low food secure), and differences between self-reported eating rate (slow, medium, or fast). Differences in eating rate, the dependent variable, between the overall food security score groups (food secure vs food insecure), the independent variables, were expected to be examined using independent t-test analyses to establish differences for hypothesis 1. Differences between quartile eating rate for quartiles 1 and 4, the dependent variables, were expected to be examined using paired t-test analyses to establish if differences existed across the meal for hypothesis 2.

Anthropometrics and Demographics
Twenty females, recruited from low-income sites, who were over the age of 18 (46.5 ± 13.7) completed the study. Of those twenty, thirteen participants were included for data analysis. Five participants from East Bay Food Pantry were excluded from data analysis because the laptop computer used to video participants did not capture the test meals correctly. As previously stated, the laptop computer was set-up prior to the arrival of participants and video recording ran continuously. Upon later review, it was determined that the five participants sat outside of the video parameter and meal duration could not be determined. Without meal duration, back-calculating eating rate was not possible and only energy intake and total grams could be calculated. Additionally, one participant refused to be video recorded and eating rate could not be back-calculated.
Lastly, one participant was excluded because the test meal duration was less than 1 minute. The participant's test meal consisted of 4 bites and the meal duration was 40 seconds. The participant was excluded because her test meal duration was an extreme outlier. The average meal duration for all 13 subjects was 5.5 ± 2.2 minutes ( Table 3).
The majority of participants were overweight or obese with BMI averaging 35 (Table 2). Due to lack of participants reporting a slow eating rate, ANOVA analysis was not conducted. There was no difference between participants who self-reported as medium or as fast eaters, those who reported as medium eaters had a similar calculated eating rate (g/min) compared to the fast eaters (t= -.807, p=0.438) ( Table 2). There was a moderate effect size between self-reported medium and fast eaters' eating rate (g/min) (d=.061).

Calculated Average Eating Rate
Statistical comparisons between food insecure and food secure groups were not possible because only one participant was in the food secure range. The calculated means for each variable are provided in Table 3.

Eating Rate Q1 vs Q4
Each test meal was split into four equal time length quartiles and quartile eating rate as bites/quartile were calculated. Bites/quartile are presented in Figure 1. There were 7 participants whose quartile 1 eating rate appeared lower than quartile 2 and 3. Within the remaining 5 participants, 3 participants decreased in eating rate and 2 participants stayed the same between Q1 and Q2. Two patterns emerged when examining the figure, those who ate with an accelerated Q4 eating rate (n=6) and those who ate with a decelerated Q4 eating rate (n=7) ( Table 4) ( Figure 2). Both the accelerated Q4 and decelerated Q4 groups had similar calculated eating rate, meal duration, energy intake, total intake, kcals/min, and bite size (Table 4). Paired t-test analysis was conducted within each of the assessed groups comparing Q1 to Q4. Within the accelerated Q4 eaters, Q1 eating rate was significantly lower than Q4 eating rate (p=.002) ( Table 5).

DISCUSSION
This was the first study to take eating rate measurement out of the lab and into a community setting. This novel methodology included covertly video recording participants onsite to calculate eating rate. Past laboratory research relied on a Universal Eating Monitor (UEM) to calculate eating rate [40][41][42] , limiting research to a lab setting. By moving research out of the lab, it opens the field to more opportunities examining eating rate in a variety of settings to further broaden research. Throughout the study several limitations and difficulties presented themselves regarding this novel methodology.
Ultimately problems with recruitment and methodology led to unexpected results and failure to answer the proposed hypotheses. However, new insights regarding this methodology were made.
The primary hypothesis was that food insecure women would exhibit a faster eating rate when compared to food secure women. The secondary hypothesis was that food insecure women would continue at a linear eating rate until meal completion whereas food secure would decrease eating rating over the course of the meal. However, neither the primary hypothesis, nor the secondary hypothesis could be tested due to several limitations and difficulties experienced throughout the study. Additionally, it is not clear whether or not the results found during this study were the result of the study population or methodology utilized during the study.
When examining the demographic data, the majority of participants were overweight or obese. Although BMI over 18.5 kg/m 2 or above was considered eligible for the study, only one participant who volunteered had a BMI categorized as normal weight.
Additionally, the majority of participants scored with very low food security. These results are supported by previous research. [43][44][45][46] Studies conducted at food pantries found that participants who frequented food pantries were severely food insecure, had higher rates of obesity, had low levels of education, and suffered from severe poverty. 43-46 Self-reported eating rate was collected and data were compared to observed calculated eating rate from the test meal. Since only one participant reported a slow eating rate her calculated eating rate could not be compared to those who reported as medium and fast eaters. Although there were no significant differences between selfreported medium and fast eaters, self-reported fast eaters did have a faster eating rate than self-reported medium eaters (Table 2). Additionally there was a moderate effect size.
Future research with a larger sample size may find significance between groups.
Past research examining calculated eating rate in comparison to self-reported eating rate was tested using a test meal energy density of 1.66kcal/g. The energy density of this study was 2.1kcal/g. Therefore results from previous studies cannot be directly compared to this study. Petty et al. examined self-reported eating rate and calculated eating rate and found that self-reported slow eaters ate at 53.0 ± 5.4 kcal/min, medium eaters ate at 63.1 ± 5.2 kcals/min, and fast eaters ate at 83.9 ± 5.5 kcals/min. 47  and rice test meal in overweight women. 48 The average eating rate was 66 ± 23 g/min, which was slightly faster than this study's 55 ± 18.3 g/min. The results are comparable and differences may be due to viscosity or texture differences between the two meals.
There is strong evidence supporting texture and viscosity differences and energy intake; findings are consistent that solids and semi-solids have stronger effects on satiety than liquids. [49][50][51][52] Additional studies include Sneddon et al. which examined eating rate in healthy college-aged women at the University of Rhode Island and served as a predecessor to this study. 8 The results found that calculated eating rate was 60.5 ± 70.75 g/min, which is slightly faster than the eating rate observed in this study. 8       Eating rate split into four equal length quartiles and bites/quartile assessed 1 Accelerated Q4 eaters = is defined as a participant who ate with a higher Q4 eating rate compared to Q1 and a higher or even eating rate compared to Q2, and Q3 eating rates 2 Decelerated Q4 eaters = is defined as a participant who ate with a lower or even Q4 eating rate compared to Q2, and Q3 eating rates Ψ Paired t-test analysis was conducted *Significance p<0.05 **p<.01 Eating rate split into four equal length quartiles and bites/quartile assessed A= Accelerated Q4 eaters, which is defined as a participant who ate with a higher Q4 eating rate compared to Q1 and a higher or even eating rate compared to Q2, and Q3 eating rates D=Decelerated Q4 eaters, which is defined as a participant who ate with a lower or even Q4 eating rate compared to Q2, and Q3 eating rates Research suggests that there is an association between obesity and lower SES, especially in women. [8][9][10][11][12] The National Health and Nutrition Examination Survey (NHANES), collected data between 2005-2008, and found that 42% of women living below 130% of the poverty line were obese in comparison to 29% of women living above 350% of the poverty line. 3 Food security also plays a role in obesity. Food insecure women are more likely to be overweight or obese than their food secure counterparts. [13][14][15] The higher rates of obesity in food insecure women may be due to poor food choices related to economic demands and lack of access to healthy food options. [13][14][15] Poor food choices such as increased consumption of energy-dense foods such as refined grains, added sugars, and added saturated and trans fat are often associated with obesity in low-income populations. 16 Several factors affecting poor diet quality are that energy-dense foods cost less, are more convenient, and are more palatable. [16][17][18][19] In addition, in low-income areas access to supermarkets can be limited. 16 Supermarkets may be several miles away and transportation may be limited or access to public transportation is poor. Low-income areas also tend to have more access to fast food and convenience stores, which may offer limited selection in fresh foods or lack healthy options. 16 To help combat food insecurity, the Federal government provides several programs for low-income populations. The Supplemental Nutrition Assistance Program (SNAP) and its education program SNAP-Ed are programs available for low-income populations. The goal of SNAP is help provide sufficient means in hopes of alleviating food insecurity and increasing access to healthy foods. 20 The goal of SNAP-Ed is to help improve diet quality by increasing nutrition knowledge. 21 An important area of research involves programs to help alleviate obesity in the U.S. Although research in the area of eating rate and obesity is far more limited; substantial evidence suggest that eating rate can also play a role in obesity. Within meals, overweight and obese individuals exhibit a faster initial eating rate than normal weight individuals, as well as do not follow the normal biological satiation curve typically exhibited by normal weight individuals. 22,23 Currently there is mixed research involving the effectiveness of eating rate interventions and reduced energy intake. [24][25][26] However, with more research eating rate interventions may become an effective tool in weight loss and obesity prevention. The limited eating rate interventions in low-income populations proved unsuccessful, suggesting that low-income populations that are food insecure differ in eating rate when compared to food secure populations. 25 To date there is no research published that examines eating rate in food insecure individuals. By examining eating rate patterns in food insecure individuals it may provide insight into creating an effective intervention for decreasing eating rate in food-insecure populations, resulting in increased weight loss and decreased obesity prevalence.

II. Obesity-Health Epidemic
What is obesity?
The World Health Organization (WHO) defines obesity as excessive or abnormal fat accumulation that poses a health risk. 27 Obesity is often the result of a lack of physical activity or exercise and the consumption of excessive calories. Food environment, genetics, and certain conditions may also play a role in obesity. Currently, the standard for characterizing obesity by The Center for Disease Control (CDC) is by BMI. A BMI over 25.0 kg/m 2 is categorized as overweight, and a BMI over 30.0 kg/m 2 is categorized as obese 28 . As the obesity epidemic continues, new classes of extreme obesity, such as BMI over 40kg/m 2 and BMI over 45kg/m 2 , have emerged. 5

Complications associated with obesity
Obesity is associated with various health-related problems such as hypertension, diabetes, dyslipidemia, certain cancers, metabolic syndrome, and cardiovascular disease among others.

Nguyen et al. examined data from NHANES between 2003 and 2004 and
established that hypertension, diabetes, dyslipidemia, and metabolic syndrome prevalence was related to obesity. 1 The results from the study found that the lowest prevalence of these health-related problems occurred in normal weight individuals, and prevalence increased as weight increased. 1 In addition, cardiovascular disease is currently the leading cause of death in the United States. 29 Poirier et al. determined that obesity play a major role in cardiovascular disease. 30 More importantly it was determined that with weight loss patients reduced their risk of CVD, thus improving overall health and decreasing the risk of arrhythmia, pulmonary hypertension, stroke, coronary artery disease, heart attack, sleep apnea, vascular disease, and congestive heart failure. 30 The relationship between obesity and type 2 diabetes is well established. For every 2.2lb (1kg) increase in weight the risk of diabetes is increased by 9%. 31 Sullivan et al. examined the relationship between obesity, physical inactivity and diabetes and determined that both obesity and physical inactivity are associated with prevalence of diabetes. 31 Flegal et al. examined obesity-related deaths in the United States. 32 After collecting data from NHANES and using follow-up data, the results showed that obesity was associated with CVD-related deaths as well as deaths related to kidney disease, diabetes, and obesity-related cancers. 32 Obesity continues to be a major health problem in the United States, and research determined that obesity plays a major role in many other major health-related issues in the United States. In addition, prevention of obesity is proven to help improve many health-related issues such as cardiovascular disease, prevalence of type 2 diabetes, certain cancers, and many other health-related issues.
Therefore, obesity prevention should be at the forefront of disease prevention and primary prevention in the United States.

III. Factors affecting obesity in women Socioeconomic Status (SES)
The causes of obesity can be simplified into two categories: environmental factors and genetic factors; however both of these categories are quite complex and can be broken down further. 7 Genetic factors do play a major role in obesity; however environmental factors are believed to contribute more to the rapid increase in obesity rates throughout the past forty years. 7 Environmental factors include socioeconomic status, diet quality, access to healthy foods, food prices, and food insecurity. 7 Other factors contributing to higher rates of obesity in the poor are growths in fast food chains, which serve energy-dense foods at low prices giving individuals maximum calories for less money. 33 In addition, advances in technology have made workplaces more sedentary, resulting in less energy expenditure. 33 Research suggests that higher rates of obesity are related to lower SES, especially in women. 11,12,16,17,19 Socioeconomic status is defined by several factors including income level, education completed, and occupation. 12 Related to education, research shows that in 2000, 26% of high school dropouts, 22% of high school graduates, and 15% of college graduates were obese. 11 In addition, 23% of women with family incomes above 400% of the poverty line were obese compared to 40% of women in low-income families. 11 Baum et al. examined the relationship between age and SES on obesity growth. 11 Data were collected from the National Longitudinal Study of Youth (NLSY), and investigated obesity throughout childhood into middle adulthood. 11 The results concluded that obesity is related to childhood SES and increases with age. 11 On average BMI is expected to increase .12kg/m 2 per year; however lower SES are predicted to have a .74kg/m 2 above their high SES counterparts. 11 In addition, for every additional year of education BMI is reduced by .20kg/m 2 ; these results were more significant for women than men. 11 The results from this study concluded that BMI is predicted to increase each year from childhood into adulthood; however, there are disparities in BMI growth and BMI is indirectly related to SES. 11 Koebnick et al. examined different populations in a cross-sectional study to determine if certain population groups were more likely to be overweight or obese. 5 California residents ages 20-39 years old were recruited and BMI, demographics, and electronic health records were collected from each participant. 5 The results found that 61.5% of young adults were overweight or obese, and Hispanics were more likely to be overweight and obese. 5 In terms of extreme obesity, women were more likely to be extremely obese and African Americans had the highest rates of extreme obesity when compared to other populations. 5 There were no differences found between men and women in obesity prevalence, only prevalence of extreme obesity. 5 Ljungvall et al. determined that there were similar levels of obesity across all income levels and determine that the obesity epidemic has affected the entire American population. 6 However differences among women's race and ethnicity and obesity, as well as differences between education level do exist. 6 Ljungvall et al. determined that African American women were more likely to obese when compared to their white and Hispanic counterparts. 6 In addition, women who did not graduate high school or with less than 12 years of schooling were more likely to be obese and severely obese. 6

Diet quality
Low-income populations often do not have the means to afford high quality foods that are low energy density. 17 A review conducted by Darmon et al. stated that low energy dense diets, high in whole grains, lean meats, fish, fresh fruits, and fresh vegetables, were typically consumed by higher SES populations. 17 On the other hand, diets high in high energy dense foods such as refined grains, added sugars, and added fats were often consumed by lower SES populations. 17 Micronutrient intakes were negatively affected in lower SES, resulting in poor diet quality; however, both macronutrient and total energy intakes were not affected by SES. 17 Darmon et al. concluded that diet quality is affected by age, sex, occupation, education level, income levels. 17 In addition, it was concluded that there is a positive relationship between SES and food quality. 17 Drewnowksi et al. examined obesity and diets with regards to social inequalities and found similar results as previously stated. 16 A direct relationship between obesity and poverty was established, especially among women. 16 Drewnowski et al. also found that energy-dense foods typically cost less than nutrient dense foods; in fact, in a survey of supermarkets in Seattle, Washington, Drewnowksi et al. established that fresh produce was ten times more expensive than vegetable oil and sugars. 16 In addition, soft drinks cost approximately 30 cents for 240 calories, whereas orange juice from concentrate cost 143 cents for 240 calories. 16 The most cost effective foods included fats, oils, refined grains, beans, and potatoes, and shelf stable foods cost far less than their perishable counterparts at the expensive of added sugars and preservatives. 16 Mello et al. examined the relationship between low-income, food insecure participants and dietary behaviors within the population. 14 The study recruited 1,874 patients from low-income health clinics, health fairs, and local social services agencies.
Participants were required to be over the age of 18, able to read Basic English, could not be pregnant, and were assessed using a Food Habit Questionnaire (FHM). 14 The FHM was scaled with low scores reflecting lower fat intakes. 14 The FHM consisted of 35 questions, on behavioral categories related to fat intake, upon which participants were scored. 14 Participants who scored with having four or less fat behavioral categories were deemed ineligible for the study for following what researchers considered a healthy diet. 14 Fruit and vegetable consumption was measured using a Food Frequency Questionnaire (FFQ); however, participants were not excluded from the study based on fruit and vegetable intake. 14 In addition to the FHM and FFQ, food insecurity, and various other measures were administered via telephone interviews with participants. 14 Low-income food insecure individuals consumed fewer vegetables and fresh fruit, had a higher-fat intake, and consumed more energy-dense foods than food secure individuals. 14 The authors suspected that energy-dense foods are perceived as having more caloric value with less cost and waste when compared to low energy-dense foods. 14 Nonetheless, the food insecurity questionnaire used had not been validated in low-income individuals and researchers believed the questions were not fully understood by participants. 14 Cortés et al. examined how nutrition education and assisted supermarket tours affect food-purchasing trends in low-income Latinos. 19 The study was a pilot study in the Boston, Massachusetts area where Spanish-speaking participants were recruited. 19 After recruitment, 20 families partook in the study. 19 Baseline demographics, food purchasing, and consumption were measured; afterwards participating families received three-five home visits for nutrition education and a supermarket tour. 19 Observations during home visits, follow-up questions, and supermarket receipts were analyzed after the study. 19 The results found that participating families significantly decreased both total calories consumed and amount of money spent at the supermarket. 19 Wolongevicz et al. examined the relationship between diet quality and obesity using data from the Framingham Nutrition Studies. 18 Participant data were collected from the Framingham Offspring and Spouse Study (FOS). 18 Participants diets' were assessed using 3-day food records and the validated global diet index, the Framingham Nutrition Risk Score (FNRS), which assigned each participant a nutritional risk score. 18 The results found women with the highest nutritional risk score were 1.76 times more likely to be overweight or obese than those who scored with a low nutritional risk score. 18 In addition, researchers found that total energy intake, fiber intake, alcohol consumption, and vitamin E intake were all negatively associated with obesity. 18 On the other hand, protein was positively associated with obesity. 18 They concluded that participants with increased nutritional risk scores had diets lower in energy, carbohydrates, and micronutrients; however, they had increased total fat intake compared to their low nutritional risk score counterparts. 18

Assess to healthy foods
A number of studies published examine the relationship between obesity and lack of access to nutrient-dense foods. Several factors behind this reasoning are lack of supermarkets in close proximity, lack of access to public transportation, and increased access to fast food and convenience store. [34][35][36] Dubowitz et al. examined food environments and used data from the Women's Health Initiative Clinical Trial (WHI). 35 Data collected from the WHI were used to assess whether or not access to supermarkets or small grocery stores and convenience stores played a role in obesity. 35 Women were recruited from both urban and suburban populations, across all race and ethnicities and a total of 60,775 women's data were analyzed. 35 The results found that there was a positive association between obesity and increased availability of convenience stores and fast food restaurants, while there was a negative association between obesity and increased availability of supermarkets. 35 Supermarkets offered a large variety of foods of all qualities, and offered more nutrientdense foods when compared to the small grocery stores and convenience stores. 35 Lastly, consumption of food from fast food restaurants was associated with increased calorie consumption, increased fat consumption, and higher BMI. 35 Lovasi et al. viewed public records to determine trends in obesity and physical activity. 36 Characteristics examined were quality and upkeep of residential and commercial buildings, transportation infrastructure, and available parks and open spaces. 36 The results found that when comparing access to a supermarket verses a small grocery store or convenience store, there was less obesity and hypertension when individuals lived in closer proximity to a supermarket and increased obesity, hypertension, and diabetes when individuals lived in closer proximity to a small grocery store or convenience store. 36 In addition, there was less obesity in areas with "walkable" areas, parks, open spaces, and access for residents to walk to food stores. 36

Eating Rate
A number of studies have examined patterns in eating rate in various populations.
Guss et al. found that eating rate, defined as number of bites per minute, measured using a Universal Eating Monitor (UEM) varied by BMI. 22 Normal BMI (18-24.9) individuals were compared to overweight and obese BMI (25-40) individuals. 22 The results concluded that normal BMI individuals followed what is considered a "biological satiation curve", meaning they started with a normal initial motivation to eat, which is defined as a rapid eating rate. 22 However, as the meal progressed the normal BMI group gradually decreased eating rate as normal inhibition and satiety senses increased. 22 The overweight and obese BMI group showed hypermotivation when compared to the normal BMI group in that the initial eating rate was considerably higher than the normal BMI group at baseline. 22 In addition the overweight and obese BMI group showed disturbed satiety, which was defined as a failure to reduce eating rate in response to inhibitory signals. 22 Zandian et al. found that overweight and obese individuals followed a linear eating rate pattern. 23 The study recruited 47 normal weight females, with a mean BMI of 22.2 from a college campus. 23 The results found that decelerated eaters initially consumed more than linear eaters, but gradually declined as the meal progressed, and linear eaters had increased overall consumption. 23 Laessle et al. examined the differences between normal weight and obese individuals in regards to initial eating rate, spoonful size, and deceleration of eating in a laboratory setting. 37 The study recruited 47 normal weight participants with a mean BMI of 22.9, and 49 obese participants with a mean BMI of 32.7 from a college campus. 37 Participants were required to fast for 10 hours prior to arriving in the lab. 37 Upon arrival, participants were given half a ham sandwich, to ensure they had the same stomach fullness. 37 Researchers controlled for stress levels, and established that there were no significant differences between stress levels. 37 Participants ate chocolate pudding, and were recorded using a UEM. 37 The results showed that obese individuals mean initial eating rate was significantly higher than their normal weight counterparts. 37 In addition, obese individuals averaged a larger portion of pudding consumed each bite; however, researchers found no difference in rate of deceleration during the progression of the meal between groups. 37

Food insecurity
What it is?
The United States Department of Agriculture (USDA) describes a range for food security. Those who report a reduction of quality, variety, or desirability with or without indications of disruptive eating patterns and reduced food intake are considered food insecure. 38,39 There are ranges for food insecurity: high food security, moderate food security, low food security, and very low food security. 40 Previously the ranges were food secure, food insecure without hunger, food insecure with moderate hunger, and food insecure with severe hunger. 39,40 Previous research may categorize food insecurity into the old categories, however this has since changed. 40,41 As of 2013, 14.3% of households in the U.S. (17.5 million households) were food insecure. 39 Approximately 5.6% of households (6.8 million households) were considered very low food insecure. 42 This means that roughly 49 million people in the U.S. are food insecure. 42 In Rhode Island 14.4% of households were considered food insecure and out of that 4.6% were very low food insecure. 42

What factors contribute to Food Insecurity?
Many believe that income is the sole contributing factor to food insecurity; however, other factors do contribute. 43 Income level is a major factor in food insecurity, with lower income populations typically more food insecure than higher income populations. 41,44 However, other factors such as time constraints can also contribute to food insecurity. 45,46 Lower education levels and little food knowledge or a reduction in self-efficacy can also play a role. [47][48][49] Additionally individuals who are either separated or divorced may be at higher risk of food insecurity. 39,46 Other factors may also include living with disability. 50 Lastly, state taxes, state wages, and cost of living can impact differences in food insecurity state by state. 51

How can Food Security be assessed?
Food security status is multifaceted, multi-staged, and complex so assessing it relies on several indicators. 40 Food conditions, experiences, and behaviors are all examined in order to determine severity. 40 Various questions in the USDA Food Security Survey Module address situations such as anxiety that households or individuals experience of not have enough food or money, experiencing running out of food, household or individual perceptions of food inadequacy, substituting with lower quality foods, and reduction in food intake to help assess food security status. Additionally the survey uses a three stage approach to help determine severity. 40 The first stage consists of experiencing anxiety that the food budget or food adequacy is inadequate and having to make modifications. The second stage involves a reduction in food intake in adults in the household, and the final stage involves a reduction in food intake in children in the household, with the adult perceiving the situation as dramatic.
The United States Department of Agriculture Economic Research Service (USDA-ERS) provides several surveys to help assess food security status. 40 The U.S.
Household Food Security Survey Module (HFSSM) is an 18-item module that assess both adult and child food security status. 40 The U.S. Adult Food Security Survey Module is a 10-item module that assesses only adult food security status. 40 The 6-item short form of the Food Security Survey Module is a quick way to assess food security status. 40

Why is it important?
Food insecurity is associated with obesity, especially in women, and other health problems such as hypertension, hyperlipidemia, diabetes, nutrient deficiency, depression, lower nutrient intakes, increased risk of birth defects, and mental health issues. 15,[52][53][54][55] Food insecure individuals, especially children, also at risk of deficiencies in iron, vitamin A, B complex vitamins, magnesium, calcium, and zinc. 15,56,57 Emerging evidence suggests a correlation between food insecurity and obesity in women. 13,44,58,59 Adams et al. examined the prevalence of food insecurity in Non-Hispanic White (NHW), African American, and Hispanic women in California.
Additionally the study examined the relationship between prevalence of food insecurity and risk of obesity. The study found that risk of obesity and prevalence of food insecurity varied among races. 13 For NHW women, the prevalence of food insecurity increased the risk of obesity. 13 However, the risk of obesity did not increase further as the severity of food insecurity increased. 13 For African American and Hispanic women the prevalence of food insecurity increased the risk of obesity and the risk of obesity increased further with increased severity of food insecurity. 13 Therefore, African American and Hispanic women who were food insecure without hunger were 1.5 times more at risk for obesity and those who were food insecure with hunger were 2.8 times more at risk for obesity. 13 Olson et al. examined the relationship between food insecurity in women of childbearing age (ages 20-39) and BMI. 60 Participants were separated into four groups: food secure (47%), household food insecure (25%), individual food insecure (17%), and households with child hunger (10%). 60 Researchers determined that the BMI's for participants who were living in "household insecure" homes, or the least severe food insecure had the highest BMI and BMIs were significantly higher than women in food secure households. 60 There were no significant differences in the more severe food insecure households (individual food insecure and households with child hunger), concluding that food insecurity was associated with increased BMI, but as the severity of food insecurity increases, BMI did not increase further. 60 Townsend et al. examined the relationship between food insecurity and BMI in over 9000 men and women. 44 The results found a relationship between food insecurity and BMI in women, but mot men. The prevalence of obesity increased with the prevalence of food insecurity in women. 44 The study found that 34% of the food secure were overweight, 41% of the mildly food insecure were overweight, 52% of the moderately food insecure were overweight, and 20% of the severely food insecure were oveweight. 44 Townsend et al. concluded that there was a significant relationship between obesity and food insecurity, with moderate food insecurity having the highest prevalence of obesity. 44 Kaiser et al. also examined the relationship between prevalence of food insecurity and prevalence of obesity. 61 Low-income Latino women were recruited and examined for food insecurity in California. 61

IV. Intervention and prevention of obesity
As the obesity epidemic in the U.S continues, prevention of obesity become vital.
Research shows that low-income populations are at a greater risk of obesity and obesity in low-income populations is associated with prevalence of food insecurity. Therefore, low-income populations are in need of programs that help alleviate food insecurity, which will hopefully allow low-income individuals make healthier decisions and have access to quality foods. The goal of SNAP is to supplement low-income populations with the means to help low-income populations improve dietary intake. 65 Programs such as SNAP-Ed and EFNEP aim to help low-income populations increase nutrition knowledge. 65 An emerging area of research involved with reducing the prevalence of obesity in food secure populations is eating rate interventions. 24-26, 66, 67 Research shows that reductions in eating rate can effectively lower BMI; however no research exists involving food insecure populations and eating rate interventions. 22,23,26 Supplemental Nutrition Assistance Program (SNAP) The largest federally funded nutrition program in the U.S is the Supplemental Nutrition Assistance Program (SNAP). 65 months. 72 In the longitudinal design those newly enrolled SNAP households were followed up at 6 months. 72 The results of the cross-over design found that 65.5% of the newly enrolled SNAP households were food insecure. 72 Out of the group of households who had already been enrolled in SNAP for 6 months, only 58.7% were food insecure. 72 In the longitudinal design, the results found that out the 65.5% of newly enrolled SNAP households that scored food insecure, only 52.8% of remained food insecure at 6 months. 72 Ratcliffe et al. found that chances of being food insecure improved by around 30% if households participated in SNAP. 73 Additionally, other research supports the notion that SNAP is an effective tool in providing adequate means to improve dietary intake to low-income populations and helps alleviate food insecurity. [72][73][74][75][76] The Expanded Food and Nutrition Program (EFNEP) is another federally funded program. 77 The goal of EFNEP is more education based, and focuses on assisting lowincome individuals in gaining the knowledge, skills, and behaviors to lead a nutritionally sound life. 77 Auld et al. examined the effectiveness of EFNEP in improving diet quality in low-income populations. 77 The results found that 95% of participants made improvements in at least one food group and 90% improved a food-related behavior. 77

Eating rate interventions
Although eating rate intervention literature is limited, it is an emerging area of research. According to the American College of Sports Medicine, self-management of meals such as eating rate are more effective in weight loss than those who do not selfmonitor. 78 Martin et al. examined whether or not slowing eating rate was associated with reduced food intake in overweight males and females. 79 Participants consumed a variety of meals consisting of various macronutrient contents, and eating rate and meal intakes were measured using a Universal Eating Monitor (UEM). 79 The results showed that when eating rate was slower, males consumed less food; however, when females reduced eating rate food intake remained the same and was not significantly different. 79 Therefore, Martin et al. concluded that decreased eating rate is a successful tool in reducing energy intake in men, but not in women. 79 Spiegel et al. examined the results of lengthening meals on weight loss in 10 obese women. 80 Participants partook in a 41-week weight control program, in which participants lengthened meal time during weeks 1-28. 80 The results found that with increased meal length there was greater weight loss; however once meal length returned to baseline during weeks 29-41 meal length was no longer associated with weight loss. 80 Spiegel et al. concluded that slowing eating rate was associated with weight loss in obese women. 80 Andrade et al. examined whether or not decreasing eating rate results in decreased energy intake. 24 Participants were recruited from a college campus, and were required to be healthy females. 24 Thirty females partook in the study, and were analyzed at two different test visits involving slowing and increasing eating rate during a test meal. 24 The results showed that there were significant decreases in energy intake during the slower eating rates and increases in meal satiation; therefore, slowing eating rate effectively reduced energy intake in women and may be an effective tool in weight management. 24

Significance of project
As stated above, obesity is a major issue plaguing the U.S, and there are many factors affecting obesity. In addition, food-insecurity is a prevalent issue in the U.S as well. Research establishes that there is an association between obesity and foodinsecurity, especially in food insecure women. 13,15,44,52,56,[60][61][62][63][64] One method for reducing obesity maybe is eating rate interventions, although much more work is needed. While short term studies have been proven to be successful in decreasing energy-intake, long term studies have not been completed. 24,80 Past research conducted at the University of Rhode Island involving eating was successful when the intervention took place on campus. 24 Students on campus typically are financially secure and are not food insecure.
However, when the eating rate intervention was moved to Providence, Rhode Island and the study population was low-income, the intervention was not successful. 25 Researchers hypothesized that because the study population was low-income, they may have been food insecure, which caused the intervention to fail. 25 Although disturbances in eating rate regulation may be related to food insecurity, there are no studies examining the relationship between eating rate and food insecurity. Therefore, the purpose of this research is to examine eating rate in food insecure populations. If there are differences in eating rate between food secure populations and food insecure populations, eating rate interventions may be tailored to food insecure populations with the hopes to reduce obesity in that population.

V. Conclusion
In conclusion, as the obesity epidemic in the U.S continues new strategies are needed to help fight it. Unfortunately, many of obese Americans are also low-income and food insecure. 13,16,41,44,60 Therefore, expensive weight loss programs are not possible for many Americans who are suffering from obesity and the many obesity-related health issues that accompany it. Inexpensive programs are needed in order to reach across all of the population and income groups. One way to help solve this is through simple eating rate interventions. In addition, past eating rate interventions have proved successful in slowing eating rate and reducing energy intake. 24,80 In fact, research shows that overweight and obese women appear to have deficits with eating rate regulation. 22,23 They exhibit an increased eating rate at baseline and either continue in a linear pattern throughout the meal or at an increased eating rate, resulting in higher total intake compared to the normal pattern of decreasing eating rate as the meal progresses. 22,23 However, there is no research examining eating rate in food insecure populations.

APPENDIX B Consent Form
You have been invited to take part in a research project described below. The researcher will explain the project to you in detail. You should feel free to ask questions. If you have more questions later, Geoffrey Greene, the person mainly responsible for this study, will discuss them with you. You can contact him at the Department of Nutrition and Food Science, 112 Ranger Hall, Kingston, RI. You must be a woman who is eligible for the Expanded Food and Nutrition Education Program (EFNEP) or Supplemental Nutrition Assistance Program-Education (SNAP-Ed), be moderately overweight based on measurement of your height and weight, be 18 -48 years old and must be fasting (no food or drinks except water) since midnight before the study and can't be allergic to eggs or milk to be in this research project or have health problems that might interfere with participation in this study.

Description of the project:
You have been asked to participate in a research study looking at appetite changes following a test meal (brunch).

What will be done:
The study will involve one visit of about one hour and fifteen minutes. At that visit, you will have the study explained and agree to participate, be weighed and measured and, if found eligible will be given a frittata for brunch and will complete questionnaires before and after the meal. You must remain in the room for one hour after starting the meal to complete questionnaires about your appetite at 20, 40, and 60 minutes after you start eating. A video recording of the test meal will be made to ensure consistency and quality control in instructions. You must be fasting (no food or drinks except water) since midnight before the meal and cannot smoke or use tobacco or engage in exercise after midnight. At the visit: • Your height, weight, and waist circumference measurements will be taken.
• You will be asked to rate your appetite on a questionnaire.
• You will eat brunch and can consume as much of the meal as you would like so that you feel comfortably full. • You will be asked to rate your appetite after finishing your meal, then 20, 40 and 60 minutes after starting the meal. You will also be asked to rate how the meal tasted. • While you are waiting to rate your appetite after the meal, you will complete additional questionnaires about your diet, physical activity, and personal and family health history. • You will receive a $20 gift card for a local supermarket as well as a free packet of nutrition information.

Risks or discomfort:
There are no known risks for the study.

Benefits of this study:
This study will help to us understand more about the relationship between food and appetite in women who are eligible for the Expanded Food and Nutrition Education program to help us develop better programs in the future. There are no direct benefits to you.

Confidentiality:
Your participation in this study is confidential. All of your information will be coded by an identification number that cannot be traced to you. None of the results of this study including the video will identify you by name. Data access will be limited to study investigators. Data will be stored in locked file cabinets and password-protected computers at the Nutrition Department of the University of Rhode Island.

Decision to quit at any time:
The decision to take part in this study is up to you. You do not have to participate. If you decide to take part in the study, you may quit at any time. Whatever you decide will in no way penalize you. If you wish to quit, you simply inform research assistant of your decision. However, you must complete the study to receive your incentive.

Rights and Complaints:
This study is part of research being conducted by the University of Rhode Island. If you have any questions or if you are not satisfied with the way this study is performed, you may discuss your complaints with Dr. Geoffrey Greene at (401)  Please sign both consent forms, keeping one for yourself.

APPENDIX C Study Test Meal Day Protocol
Preparations before subject's arrival: Research Assistants (RA) should prepare folders with all questionnaires labeled with the subject number and two copies of the consent forms. RAs should arrive at the location one hour before the first appointment. This time will be spent cleaning the room and all surfaces (using antibacterial all-purpose cleaner) that will be used during the meal and meal preparation.  Responses of "yes," "often," "sometimes," "almost every month," and "some months but not every month" are coded as affirmative. The sum of affirmative responses to the 10 questions in the Adult Food Security Scale is the household's raw score on the scale.
Food security status is assigned as follows: • Raw score zero-High food security among adults • Raw score 1-2-Marginal food security among adults • Raw score 3-5-Low food security among adults • Raw score 6-10-Very low food security among adults For some reporting purposes, the food security status of the first two categories in combination is described as food secure and the latter two as food insecure.
(2) Response Options: For interviewer-administered surveys, DK ("don't know") and "Refused" are blind responses-that is, they are not presented as response options but marked if volunteered. For self-administered surveys, "don't know" is presented as a response option.
(3) Screening: The two levels of screening for adult-referenced questions are provided for surveys in which it is considered important to reduce respondent burden. In pilot surveys intended to validate the module in a new cultural, linguistic, or survey context, screening should be avoided if possible and all questions should be administered to all respondents.
To further reduce burden for higher income respondents, a preliminary screener may be constructed using question HH1 along with a household income measure. Households with income above twice the poverty threshold AND who respond <1> to question HH1 may be skipped to the end of the module and classified as food secure. Using this preliminary screener reduces total burden in a survey with many higher income households, and the cost, in terms of accuracy in identifying food-insecure households, is not great. However, research has shown that a small proportion of the higher income households screened out by this procedure will register food insecurity if administered APPENDIX K