, War .5. : V .3 33.3.! . my? mum... V .. {mg .3. Mg? m. ‘ »I.. u. us; . film . .2": . _ A . . . ‘ , . ,. _. i 3w. x 3.. A . . ‘ , . , . . . . at... .. . . fax, a! .. . ‘ . «3%: f: ; . h .. ... z»... m. L_. ..\ THEQ’ ., V‘s. "n :2 cm This is to certify that the thesis entitled Eating habits, physical activity and risky behaviors of youth practicing weight control. presented by Julie Lynn Chmielewski has been accepted towards fulfillment of the requirements for Masters degree in Science ‘seiégifif1_,.SL4¥AD¢AA_R25DTFN§E> Dr. Sharon Hoerr Major professor ‘3‘- “4" 00 D3“: December 15.2000 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE a DflE QUE \ o _‘ v V 3"” mi/ ”21.57200; APR 0 4 tilt; 6/01 c-JCIRCJDateDue.p65-p.15 EATING HABITS, PHYSICAL ACTIVITY AND RISKY BEHAVIORS OF YOUTH PRACTICIN G WEIGHT CONTROL By Julie Lynn Chmielewski A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Food Science and Human Nutrition 2000 ABSTRACT EATING HABITS, PHYSICAL ACTIVITY AND RISKY BEHAVIORS OF YOUTH PRACTICING WEIGHT CONTROL By Julie Lynn Chmielewski This secondary data analysis was conducted to: 1) evaluate a nationally representative sample of adolescents to determine the prevalence of weight control methods according to youth’s weight status and 2) examine if overweight youth are using the recommended methods of weight control according to Healthy People 2010. The data were from the National Longitudinal Study of Adolescent Health of pubescent adolescents (N=13,570) ages 11 t018 yr. Health behaviors were examined by weight categories of under- (UW), average (AW) and overweight (OW) using Body Mass Index (kg/m2). Analyses showed youth of all weight categories used some form of weight control; the highest percentage were OW youth (89%), then AW youth (55%), and even UW youth (30%). AW youth used the most extreme methods of weight control e. g. laxatives, vomiting and diet pills. The adolescents most likely to use weight control were young, white and female. Boys and girls most frequently reported the use of “exercise for weight control” (45% & 52%). Weight control was associated with positive health behaviors of increased exercise (OR=1.59) and eating less fast food (OR=1.54), but with the negative behaviors of skipping breakfast (OR=1.56) and consuming more alcohol (OR=1.58). Those using weight control were not more likely to eat five fruits and vegetables, unfortunately, or to drink milk once a day or to eat a variety of food. The findings suggest the importance of assessing weight control use in all adolescents to ensure that health is not being compromised. Copyright by JULIE LYNN CHMIELEWSKI 2000 ACKNOWLEDGEMENTS To the members of my Guidance Committee: The guidance each member gave towards putting this thesis together was invaluable. Each took a unique role in supporting both my professional and personal life, and I thank them dearly. To Dr. Sharon Hoerr, Department of Food Science and Human Nutrition, who served as my major advisor, her desire to learn everything she could about the field of nutrition and to pass this knowledge on to me during my graduate career was invaluable. To Dr. Lorraine Weatherspoon, Department of Food Science and Human Nutrition, who not only served on my committee but was also a true friend. Thank you for having confidence in my skills and teaching me the value of balancing persistence and patience. To Dr. David Kallen, Department of Pediatrics and Human Development, who served as my editorial taskmaster© Thank you for your timely feedback, even during your “retirement days”. Your skills with the written word are astounding and give me something to strive for. To my lab group, Michelle, Anna, Deb and Seung-yeon, for their constant support. Thank you for listening to all of my “practice” lectures and making graduate school more than just an academic experience. Also, a thank you to my adopted lab group, Alison, Deepa, and Tara, for their encouragement. To Dr. Wanda Chenoweth, Mrs. Stella Cash, Dr. Jenny Bond, Diane Golzinski and Debie Lecato, who each played an important role in my graduate career, and made my success possible. iv To Jim, thank you for your friendship, tolerance and knowing just what to say when I needed it most. Your encouragement during the good and “not so good” times will not be forgotten. Finally, thank you to my parents, Norman and Claudia, who stuck by me through all of my career changes and encouraged me to do my best. Your wisdom and support cannot be described in a few short sentences but know that I appreciated ALL of it. LBT' lJSTi CHAF INIRI "l 0.. CHAP REV” Sd t; PM We (HAP “LII: Dr Rn Dd Am CHIP RESt] CHAP' DISCt CHAP Sixth APPE) REFEF TABLE OF CONTENTS LIST OF TABLES ................................................................................ vii LIST OF FIGURES ............................................................................. ix CHAPTER 1 INTRODUCTION .................................................................................. 1 Research Questions .............................................................................. 7 Hypotheses ........................................................................................ 8 Glossary ........................................................................................... 9 CHAPTER 2 REVIEW OF THE LITERATURE ............................................................. 11 Selected adolescent health databases .......................................................... 11 Weight control in adolescents .................................................................. 17 Puberty of adolescents .......................................................................... .26 Weight categories of adolescents ............................................................... 28 CHAPTER 3 METHODS ........................................................................................... 31 Data source & participants ..................................................................... 31 Sampling and weighting ........................................................................ 32 Research Instrument ............................................................................ 36 Definition of variables ........................................................................... 36 Analyses ............................................................................................. 44 CHAPTER 4 RESULTS ............................................................................................ 48 CHAPTER 5 DISCUSSION ........................................................................................ 67 CHAPTER 6 SUMMARY AND IMPLICATIONS .......................................................... 75 APPENDICES ...................................................................................... 80 REFERENCES .................................................................................... 85 vi 2.] 2.2 2.3 2.4 2.5 3.1 3.2 3.3 3.4 3.5 4.1 4.2 4.3 LIST OF TABLES Large, recent databases of adolescent health Recent survey of research on weight control behaviors of adolescents Grth patterns and body composition changes in boys and girls. Sexual maturation stages for boys and girls. Weight categories of youth as defined by the new NCHS growth charts. Selected foods categorized into five food groups from Add Health Wave II. A comparison of pubescent girl’s measured versus self-reported heights and weights used to calculate BMI from Wave II weighted data (n=5788). A comparison of pubescent boy’s measured versus self-reported heights and weights used to calculate BMI from Wave II weighted data (n=4479). How puberty status was determined for boys in Add Health Wave II. A description of analyses run for each hypothesis Demographic characteristics for pubescent girls and boys participating in Wave II of Add Health. Crosstabulations of weight control uses by the variable “Was a third person present in the room during the interview?” Crosstabulation of weight control methods of girls by the variable “Was a third person present in the room during the interview?” vii Page 12 18 26 28 29 38 42 42 44 47 48 51 52 4.4 4.5 lb 48 4.9 4.10 5.1 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 5.1 5.2 Crosstabulation of weight control methods of boys by the variable “Was a third person present in the room during the interview?” Weighted percentage of weight control methods used by girls and boys in Wave II of Add Health. Logistic regression analyses of demographic variables by weight control for girls and boys participating in Wave II of Add Health. A comparison of risky health behaviors for girls using and not using weight control. A comparison of risky health behaviors for boys using and not using weight control by weight category. Logistic Regression analyses of healthy behaviors of girls using weight control as reported in Add Health Wave II. Logistic Regression analyses of healthy behaviors of boys using weight control as reported by Add Health Wave II (n=1299). Logistic regression analyses of use of health behaviors by average and overweight boys and girls. Are certain weight control methods associated with risky health behaviors in girls? Are certain weight control methods associated with risky health behaviors in boys? A comparison of the prevalence of weight control in adolescents of three adolescent health databases. Comparison of weight control methods of youth from three different surveys. viii 52 54 56 58 58 60 62 63 65 66 67 70 a J LIST OF FIGURES Page 3.1 Sampling information from the Add Health data 32 3.2 Selection of subjects based on specific criteria in Wave II 35 of Add Health (N=11,414) ix .L'T IICIf INTRODUCTION The weight control methods used by youth in our society are raising concerns among health professionals and parents. The increase in use of weight control is due in large part to the increasing prevalence of obesity in all age groups, the media portrayal of a thin ideal body image and myriad of “quick fixes” available to consumers to help them emulate this manufactured idea of the ideal body. Adolescents fall prey to these “quick fixes, and this may be contributing to the high prevalence of inappropriate weight control practices in this age group. In addition, studies show dieting and weight control tends to begin during adolescence (Richards, Casper & Larson, 1990). Many studies to date have centered on weight control among young adults, omitting the critical group of adolescents. The current lack of adolescent studies may be due, in part, because it is difficult to involve adolescents directly without permission from a parent or guardian. Secondary data analysis of adolescent national databases is one way to investigate youth and adolescent health behaviors. The emphasis on thinness in contemporary America is thought to play a key role in the development and maintenance of eating disorders (N eumark-Sztainer, Story, Dixon & Murray, 1998a; Stettler, 1999), and weight concern and dieting have become a social norm for youth (Huon & Strong, 1996). In fact, there are more adolescents trying to lose weight than there are actually overweight (CDC, 2000). The 1997 Youth Risk Behavior Surveillance (YRBS) indicated over one-third of the youths surveyed were, “dieting to lose weight”, or “dieting to keep from gaining weight”, and girls were more likely to use weight control than boys were (CDC, 2000). Over 10% of these girls and 4% of these boys tried to lose or maintain weight through dangerous methods such as use of diet pills, lift} fixer me: hfiai COu' laxatives or vomiting (Kann, Kinchen, Williams, Ross, Lowry, Hill, Grunbaum, Blumson, Collins & Kolbe, 1997). There are a wide variety of diet pills and weight loss gimmicks on the market today; some have been proven harmful and others not harmful. Diet pills are partially considered a dangerous form of weight control in adolescents, because past research findings have associated adolescent diet pill use with more harmful drug use such as amphetamines (Gritz & Crane, 1991). Amphetamines can cause neurological complications such as seizures, stroke and psychic alterations (Dietz, Tejedor, Tejada & Frank, 1989). Also, different forms of diet pills may be harmful themselves or may take the place of meals in an adolescent’s diet. Extreme weight control has serious mental and physical health consequences for youth. Growth retardation may result from prepubescent children who intentionally restrict their nutrition intake and their fat intake (Pugliese, Lifshitz, Grad & Marks—Katz, 1983; Lifshitz & Moses, 1989). Another problem for youth are those with parents who restrict their diets, particularly fat intake, because the increased awareness of adolescent obesity has frightened the parents. Other physical health problems associated with excessive weight control behaviors include acute gastric dilation and/or rupture, metabolic alkalosis, cardiac arrhythmia, or even death (Huon & Strong, 1996). Mental health consequences resulting from extreme weight control involve depression, which could be associated with suicide (Neumark-Sztainer et al. 1998a; Richards et al., 1990), smoking initiation (Tomeo, Field, Berkey, Colditz & Frazier, 1999), substance abuse and development of eating disorders (N eumark-Sztainer et al. 1998a). Youth with eating disorders may intentionally restrict their food intake out of fear of weight gain (Levine, Smolak, Moodey, Shuman & Hessen, 1994b; Pugliese et al., 1983). 0'. fro\ Allan (1998) reported that twenty percent of all children and adolescents are overweight as defined by a Body Mass Index (kg/m2) exceeding the 85th percentile of the old National Center for Health Statistics (NCHS) growth charts. Body weight is a function of genetic predisposition, stage of physical maturity, and partially a result of health behaviors such as eating and activity patterns (Garn, LaVelle & Piklinton, 1984). Healthy guidelines state that a balance of eating and exercise will aid in decreasing the prevalence of obesity in adolescents. However, these guidelines are not always interpreted correctly and this may be a factor contributing to eating disorders and dangerous weight control behaviors in adolescents. There are some research studies on the cause of dieting in adolescents, but a scarcity of data on how dieting and weight control impact eating and physical activity behaviors in youth. In the only study located to date, the impact of weight control methods on dietary intake and on physical activity was examined. Adolescents in this study were divided into three groups signifying extreme, moderate or no weight loss methods practiced (Story, Neumark-Sztainer, Sherwood, Stang & Murray, 1998). The researchers used self-reported data from the cross-sectional 1993 Youth Risk Behavior Survey (YRBS) on 16,296 adolescents in grades 9 through 12. Extreme weight control included adolescents using vomiting or diet pills to lose weight. Health-promoting behaviors were divided into two categories of dietary and physical activity behaviors. Diet—related health-promoting behaviors included eating the appropriate five fruits and/or vegetables to meet the “F ive-A-Day” requirements and/or eating less than two high fat foods during the previous day. Physical activity health-promoting behaviors included some exercise seven days a week and vigorous exercise four or more days a week. Story £07151 start in It Wei C00 et al., (1998) showed extreme dieters were significantly less likely to eat fruits and vegetables and were more likely to have consumed two or more servings of high—fat foods during the previous day than moderate dieters and non-dieters. Moderate dieters consumed a diet more healthful than even the non-weight controllers did. Some limitations are apparent in this study by Story et a1 (1998). Researchers categorized some individuals as moderate dieters, even if they used laxatives. Also, the questions were self-reported in a setting with peers present, which may have led to over or under- reporting of both weight control behaviors and health-promoting behaviors. Pubertal status of individuals was not considered, and adolescent body weight was not controlled in the analyses. It is important to control for weight, because current studies show that weight is partially responsible for weight control behaviors practiced by adolescents. One method now recommended for use to assess overweight in adolescents is to compare a calculated Body Mass Index (BMI) of weight in kilograms divided by height in meters squared to the new NCHS grth charts, which can be found at http://www.ch.gov/nchs/about/major/nhanes/growthcharts/clinical_charts.htm (Dietz, 1998). Health care providers and nutritionists sometimes use a combination of more appropriate measures than BMI to assess obesity in adolescents. Such measures include body fat and fitness. Measures to assess body fat are bioelectrical impedance and/or skinfold thickness (Sardinha, Going, Teixeira & Lohman , 1999; Malina & Katzmarzyk, 1999). When available, a combination of measures is preferred to BMI alone, because BMI fails to consider the sexual maturation stage of an adolescent Walcin & Kinik, 1999; Daniels, Khoury & Morrison, 1997), race (Daniels et al., 1997), growth stunting (Schroeder & Martorell, 1999) muscle mass (Garn, Leonard & Hawthron, 1986) or bone “cf 92 \u. Such lDIEr ) inf» ”r density. BMI alone is appropriate to describe a population, but not to clinically diagnose an individual adolescent as overweight. However, health professionals often resort to using BMI alone, because BMI is routinely obtained in the clinical setting, is relatively inexpensive and is easy to measure (Dietz, 1998). In this study, data were used to examine adolescents (N =1 1,414) practicing weight control from Wave II of the Longitudinal Study of Adolescent Health (Add Health) out of the Carolina Population Center at the University of North Carolina at Chapel Hill. Also, determined was how BMI of these adolescents relate to health behavior patterns of eating, physical activity and risky behaviors such as smoking, drug use, alcohol consumption and/or suicide. Because family influences may have an impact on what and how an adolescent cats and exercises, variables concerning the family were also addressed. The study was unique in that: 1) the sample was representative of non- institutionalized US adolescents enrolled in regular schools, 2) factors affecting weight such as puberty status were determined, 3) heights and weights were measured by trained interviewers in the targeted group rather than self-reported, 4) detailed nutrition information was collected on food items eaten, meals eaten, beverages consumed and vitamin/mineral supplementation, 5) family influences on eating are present in the variables of Add Health. Adolescents may interpret “weight loss” terminology negatively. This could potentially lead to inappropriate health behaviors. Adolescence is a time of radical physical and mental growth and social changes. Such changes are difficult for many youth to handle, and this difficulty is manifested in a number of ways, some of which result in the use of dangerous weight control practices and poor eating habits. Results of :3 \_ 33 I Ix. weight control practices may be deterioration in physical and mental development, which could potentially lead to serious long-term health consequences. This study may provide useful information for professionals who counsel adolescents on appropriate weight status and health behaviors. L’J 2. RESEARCH QUESTIONS a. What is the frequency of weight control use among adolescents who have already started puberty and are grouped as underweight, average weight or overweight lobese using measured heights and weights to calculate Body Mass Index? b. Are there differences in frequency of weight control use among pubescent youth by weight category, grade, race, and age category? c. Of the pubescent youth using weight control, what weight control methods are most frequently used in the weight categories of underweight, average weight and overweight/obese? Weight control methods include dieting, exercise, use of diet pills, laxatives and vomiting. a. Are pubescent youth, who use weight control methods, at an increased risk for participating in other risky health behaviors such as cigarette smoking, drug use, alcohol consumption and/or suicide attempts compared to those who do not use weight control? b. Is there a difference in risky behaviors among pubescent youth using weight control and grouped by underweight, average weight and overweight/obese? Do underweight, average weight and overweight pubescent youth who use weight control have healthy eating and exercise behaviors? HYPOTHES ES Pubescent youth using weight control are more likely to be overweight/obese, White and younger in age. Pubescent youth who use dangerous and non-dangerous weight control methods are at a higher risk for participating in risky health behaviors of tobacco use, drug use, alcohol consumption and suicide thoughts than those pubescent adolescents who do not use weight control. Underweight, average weight and overweight/obese youth are more likely to use risky health behaviors when they are using weight control behaviors. Youth of any weight category using weight control have unhealthy eating/exercise behaviors as depicted by a lack of frequent exercise, a decreased consumption of fi'uits, vegetables, milk and breakfast, and an increased consumption of fastfood. U) A\ e EIQl and Dan S€\ fro: He (1m €11 mt - Ht Iiil 0i PL in! lilt Se nu GLOSSARY . Average weight = Based on the National Center for Health Statistics (NCHS 2000) growth charts, a BMI above the 5th percentile and below the 85‘h percentile by age and gender. . Dangerous weight control = Based on any of the following methods in the previous seven days including vomiting, laxatives and use of diet pills to lose weight or keep from gaining weight. . Healthy eating = Eating a variety of foods from all food groups, consuming fruits and vegetables at least five times a day, consuming at least one glass of milk daily, eating fast food less than two times a week and consuming all food types in moderation. . Healthy exercise = Participating in exercise three to five times a week for at least thirty minutes. . Overweight/obese = Based on the NCHS 2000 growth charts, a BMI above the 85‘h percentile by age and gender. . Pubescent boys = Boys who have reached a level or three or higher on the sexual maturation stages as defined by self-reported secondary sex characteristics. . Pubescent girls = girls who have started menstruation prior to the Add Health interview. . Risky health behaviors = Any behavior that puts the health of an individual at risk including smoking, alcohol use, suicide and drug use. . Sexual maturation stage = the stage of sexual maturation for youth defined by a number of secondary sex characteristics previously researched by Tanner (1984). 10. Underweight = Based on the NCHS 2000 growth charts, a BMI below the 5‘h percentile by age and gender 11. Weight category = Using BMI measures and the NCHS 2000 growth charts to categorize adolescents by underweight, average weight and overweight or obese. 12. Weight Control Methods (Non-dangerous and dangerous)= using any of the following methods in the previous seven days including dieting or restrictive eating, exercising, use of diet pills, laxatives, or vomiting to try and lose weight or keep from gaining weight 10 LITERATURE REVIEW The literature reviewed includes a comparison of adolescent health databases to the Longitudinal Study of Adolescent Health (Add Health). A general description of weight control and the health consequences of and health behaviors associated with weight control in adolescents are also examined. Finally, reviewed are puberty and weight category as they relate to weight control. Selected adolescent health databases Adolescent health surveys are collected periodically and some are nationally representative of youth in the United States. Such local, regional and national surveys of adolescent health issues are an important resource for research, because conducting research on adolescents can involve high costs and parental/ guardian consent. Table 2.1 gives a brief description of sample research foci of the large adolescent health surveys within the last five years. These databases were selected from searches on the US Department of Commerce Technology Administration (http://www.ntis.gov/index.html), Combined Health Information Databases (CHID) web site (http://chid.nih.gov/subfile/subfile.html) and Michigan State University Medline using the key variables adolescent, health, survey and database. Each database has unique components, which will be compared to the Add Health database. These databases are comparable to Add Health in that they are large samples of adolescents, they ask questions on health behaviors and/or are nationally representative of non-institutionalized US adolescents. 11 l EFP'Fl o. f; :"T‘ ' v o (fie-r" ’. 71‘ O I -..l v—A (I) '/ Table 2.] Large, recent databases of Adolescent Health Database Demiographics CateLories of Interest Longitudinal Study of n = 20,745 I eating habits Adolescent Health (Add Ages 11-23 yr I physical activity Health) Ethnicity I tobacco use http://www.cpc.unc.edu/proje Gender I substance abuse cts/addhealth National I measured ht. & wt. Youth Risk Behavior n= 16,262 I weight control Surveillance (YRBS) Ages 12-21 yr I tobacco use http://www.cdc.gov/ncsh/abo Ethnicity I alcohol ut/about.htm Gender I drug use National I physical activity I self-reported ht. & wt. Minnesota Adolescent Health n=30,000 I health status Survey (MAHS) Ages 12-20 yr . . . . I health behavrors. Ethnrcrty use of tobacco, alcohol, Gender . . . . . illrcrt drug use, su1c1de SES attem ts wei ht control Self-reported height _ p ’ . g and weight psychosocral factors I self-reported ht. & wt. State National Longitudinal n = 9,000 I self-reported onset of Surveys of Youth Ages 12-16 yr puberty 1997(NLSY97) National I cigarettes http://stats.bls.gov/NLSY97. I drugs htrn I alcohol I self-reported ht. & wt. National Health and 11: 7,344 _ tobacco use Nutrition Examination Ages 2-16 yr . . . . I rllrcrt drug use Survey (NHANES) National . . . I surcrde thoughts http.//www.cdc.gov/nchsww . . . I eatmg behavrors w/about/major/nhanes/hanesg I measured ht. & wt. ols.htm The Longitudinal Study of Adolescent Health (Add Health) is a nationally representative database of non-institutionalized adolescents of the US population and was developed by researchers at the University of North Carolina at Chapel Hill, North Carolina (Bearrnan, Jones & Udry, 1998). Over 20,000 adolescents were interviewed during 2 periods, Wave I and Wave II of the survey, from 1994 to 1996, in both a school 12 the r Ii 1411 infill and home setting respectively. The results were divided up into public- and private-use databases. Health behaviors of the database include diet, physical activity, health service use, injury, violence, sexual behavior and substance abuse to name a few. The National Institute of Child Health and Human Development and 17 other federal agencies funded the project. The public-use data set can be requested at the web address http://www.cpc.unc.edu/projects/addhealthl. To obtain the private-use data set, which doubles the sample size, researchers must follow rigorous security procedures for access. Unlike Wave I, Wave II questions were modified to contain more detailed nutrition information, risky health behaviors and measured heights and weights for the adolescents along with the self-reported heights and weights. The same adolescent sample was used for both waves of the interviews with the exception of those seniors who graduated from Wave 1. A slight disadvantage of using Add Health for this study is that some sensitive weight control questions were not asked in a private setting. The lack of privacy creates the possibility for under-reporting of certain behaviors. The limitation, however, is more than countered by the advantage of using a large data set permitting the ability to examine influences on youth in the US and the cost saving from using already collected data. The Youth Risk Behavior Surveillance System (YRBSS) was developed by the Centers for Disease Control (CDC) to monitor six categories of behaviors that contribute to the leading causes of morbidity and mortality of US youth (Kann et al., 1998). The categories from YRBSS that relate to this study include tobacco use, alcohol, other drug use, unhealthy dietary behaviors and physical inactivity. The YRBSS includes a “patchwor ” of national, state, territorial and local, school-based surveys of high school 13 students. Surveys are conducted on an annual basis and results are reported on the YRBSS web site at http://www.cdc.gov/ncsh/about/about.htm. The original survey is a convenience sample, but when weights are applied to the database, the information becomes nationally representative of non-institutionalized US students grades 9-12. Anonymous and voluntary participation protects students’ privacy. Students complete the self-administered questionnaires in classrooms during regular class periods, often during physical education and health classes. The core questionnaire contains 87 multiple-choice questions, but there are no reports for how many minutes it takes to complete the questionnaire. Local parental permission procedures were followed prior to survey administration. Limitations of using the YRBSS over Add Health include data collection technique, possible inaccurate self-reports of height and weight by growing children, and the lack of detailed nutrition questions. Distributing questionnaires in a classroom is inexpensive and convenient, because students are readily available and surveys can be collected immediately. This, however, allows for student interaction or teacher influence while filling out a survey. Peer pressure to select certain answers may arise, resulting in over-reporting on some questions and under-reporting on others, especially for sensitive issues such as suicide attempts, weight control by vomiting or laxative use and substance use. In addition, much of this sensitive information may be omitted due to the lack of privacy in the testing environment. By comparison, Add Health variables of interest were collect via interviews in the respondent’s home, and sensitive questions, with the exception of weight control, were asked through headphones and entered directly into a computer without the parent’s l4 Pill lid: and knowledge. Add Health, Wave II, also has more detailed questions than YRBS on type of foods consumed, meal consumption and beverage consumption. Also, trained interviewers measured the heights and weights of the adolescents. In 1986-1987, approximately 30,000 adolescents in the Minnesota public schools completed the Minnesota Adolescent Health Survey (MAHS) to assess adolescent health status, health behaviors and health concerns. Food intake patterns, weight control practices and disordered eating behaviors are some variables of interest. The data for grades 7 through 12 were collected in classroom settings. While this is a good data set to study adolescent nutrition, it is not representative of the US population, unlike the Add Health database. Also, the MAHS questions were asked and answered in a non-private setting, and heights and weights were self-reported. Data from the MAHS can be used to study behaviors reported by adolescents in Minnesota, but the information cannot be generalized to the US adolescent population. The Bureau of Labor Statistics, US Department of Labor, Sponsors the National Longitudinal Surveys of Youth 1997(NLSY97) (Bureau of Labor Statistics, 1999). The purpose of the NLSY survey is to document the transition from school to work on a nationally representative sample of 9,000 US adolescents ages 12-16 years . The data were collected from hour-long personal interviews on both youth and parents of youth. The NLSY asks a variety of questions on adolescent risky behaviors such as tobacco, drug and alcohol use. The data set also contains information on pubertal status of youth, which is self-reported. The fact that the NLSY does not include questions on nutrition and weight control methods is a limitation of its use for those research questions. The 15 pLW‘SL .p\ A .,l 3 1 “0-4.... ITEOI‘LilI. for H: Em. purpose of NLSY was work behaviors, not health behaviors, of adolescents, therefore one would not expect to see these behavioral questions. The Third National Health and Nutrition Examination Survey (NHANES 111), 1988-1994 is a periodic survey conducted by the Centers for Disease Control’s National Center for Health Statistics (http://www.cdc.gov/nchS/nhanes.htm) and designed to monitor factors affecting the health and morbidity of the US population (National Center for Health Statistics, 1996). Information was collected by older adults using the Mobile Examination Centers (MEC). The MECS provided a private setting for more sensitive questions. NHANES III includes a nationally representative sample of children in the US with data on 7,344 persons ages 2 months to 17 years. NHANES III is comparable to the Add Health database, because weight and height of the adolescents is measured and sexual maturation stage is assessed during a physical examination. Sensitive questions on tobacco, drug use and alcohol were also asked and answered to protect the respondent privacy through the use of headphones. NHANES 111 questions cannot be used to determine the detailed weight control methods of all youth, because specific weight control methods were only asked on a sub-sample of adolescents (n~2000). Add Health was conducted in a more private setting when youth answered questions and produced more detailed nutrition questions than YRBSS or MAHS. NLYS97 has an inadequate amount of questions on health behaviors and nutrition intake of the youth in comparison to Add Health. Although the dietary data from Add Health were less deatiled than those from NHANES 111, Add health has questions to examine specific weight control methods on a larger sample of youth. The Add health database contains responses to questions about other risky health behaviors, different weight 16 control met weight and ‘ My St“. national dd'r summit 0'. control. as '. Weight c012: bvih ollhcs control methods, detailed nutrition information and trained interviewer measurements of weight and height to answer the specific research questions of the study. Weight control in adolescents Several researchers have studied weight control behaviors of adolescents using national databases (CDC, 2000; Story, French, Resnick & Blum, 1995; Strauss, 1999). A summary of these studies can be found in Table 2.2. Different methods of weight control, as well as, reasons for use of weight control were examined. Studies focused on weight control use by gender and by ethnic groups to see if any differences existed within both of these categories. 17 Tabh Atmor Tomb Sane} iCDC. Bh.ne Heidi {Star} &l%ar \ NHA\ (Shun Table 2.2 Recent survey of research on weight control behaviors of adolescents Author Sample size Findings Youth Risk Behavior 16,000 boys and I 43% trying to lose weight Survey — 1999 girls I 61% females (CDC,2000) Ages 12-21 yr I 25% males I 4% laxative use I 30% dieting I 8% diet pills I 59% exercised Minnesota Adolescent 36,320 boys and Weight control behaviors: Health Survey girls I males 8-15% (Story, French, Resnick Ages 12-20 yr I females 27-43% & Blum, 1995) Laxative and diuretic use: I 2% for all ethnic groups Overweight/obese by BMI: I 31% White I 38% African American I 40% Hispanic I 40% American Indian I 23% Asian NHANES 1,932 boys and I 24% overweight (Strauss, 1999) girls I 10% obese Ages 12-16 yr I No significant racial differences in dieting No % given for youth using weight control National surveys have shown that dieting and attempts at weight control most frequently begin during adolescence (Richards et al., 1990). The comprehensive YRBS in 1997 found 30% of youth reported dieting to lose weight or to keep from gaining weight (Kann et al., 1998). Over 10% of girls and 4% of boys tried to control their weight through dangerous methods such as use of diet pills, laxatives or vomiting (Kann et al., 1998). The percentage of youth using weight control has increased in the most recent results of the YRBS from 40% of all youth in 1997 using weight control to 43% in 1999 (CDC, 2000). Risky behavior reported in the YRBS included a serious 18 c ri' ‘ Lula r\ “ CFC Were 85:] I) at ., ‘900 consideration of attempting suicide (21%), trying cigarette smoking (70%), regular tobacco use (17%), frequent alcohol use (33%), trial of marijuana (47%), and trial of cocaine (8%). No published reports have been located on the association of weight control with these risky health behaviors. The Minnesota Adolescent Health Survey, conducted in 1995 on about 17,000 girls and 16,000 boys grades 7 to 12, showed weight behavior patterns and that unhealthy weight control behaviors were not confined to upper socioeconomic status, White females (Story et al., 1995). Researchers reported weight control behaviors were less common in males (8-15%) as compared to females (27-43%). Laxative and diuretic use for weight loss was about 2% for all ethnic groups. Black females were significantly less likely to have dieted frequently in the past year (odds ratio [OR]=0.59) and less likely to view themselves as overweight (OR=O.41) when compared to white females. Asian males were more likely to report binge eating (OR=1.31) compared to white males. Black males were more satisfied with body weight (OR=1.57). Richard Strauss (1999) reported that 24% of youth aged 12-16 years in 1994 were overweight and 10% were obese based on analyses conducted on the NHANES III database. Of youth who considered themselves overweight, 42% actually were of average weight when their weight for height was compared to growth chart norms from the National Center for Health Statistics NHANES I and II data (F risancho, 1990). Furthermore, girls were significantly more likely to misclassify their weight category than boys (52% and 25% respectively). White girls who thought they were overweight were three times as likely to have a BMI less than the 85‘h percentilel compared with ‘85“ percentile — defined as overweight by the 2000 NCHS growth charts. See NCHS 2000 in the appendices. l9 \n'h ' klhll dd‘ “ 57': OK white boys (OR=3.5), African American boys (OR=3.6) and black girls (OR=3.1). There were no reported differences in self-perceived weight category among African American and White boys. Adolescent White girls were more likely to diet than Afiican American girls, African American boys and White boys. Dieting and weight control was associated with adolescents viewing themselves as overweight, independent of whether they actually were overweight. The adolescent databases reviewed Show from 8-23% of boys and 27-60% of girls used weight control. The method of data collection and the under- or over-reporting of health behaviors by adolescents may explain the large range in these percentages. Weight control methods reported by youth included dieting or exercising to lose weight or keep from gaining weight, as well as, the use of dangerous methods such as diet pills, laxatives and vomiting to lose weight or keep from gaining weight. After determining the percentage of youth using weight control in the Add Health database, other relationships such as weight classification, ethnicity, age, grade and health behaviors can be explored to determine individual characteristics of adolescents using weight control. Kilpatrick, Ohannessian & Bartholomew (1999) investigated the weight measurement activities and weight perceptions among adolescents using the Add Health database. They examined the relations of school health education to these activities and perceptions by youth. The researchers used the sample of 6,500 adolescents grades 7-12 from Wave I of the public use database, and questions from the in-home interviews and in-school questionnaires. Kilpatrick et al. reported 20% of adolescents trying to gain weight and 33% attempting to lose weight with females more likely than males to be attempting weight loss (47% vs. 20%) and males more likely to attempt weight gain (31% 20 VS. 709l- -\ attempt we: relate to \i 160° ti and Air Slglllflillil p5‘h percentile & < 85‘h percentile Overweight 2 85th percentile 7 < 95th percentile Obese 3 95‘h percentile * Age and gender specific. BMI is a convenient, public health indicator of weight trends, because BMI requires only measures of weight and height and reduces both measures to one number comparable over a range of heights. The interpretation of BMI, however, is based on the assumption that height is stable, obviously not true for growing adolescents whose gr OWth spurts during maturation are preceded by increases in body fatness (Eveluth & Tanner, 1990). Furthermore, children maturing earlier than average would be identified as Obese by the 95‘h percentile of BMI for age and gender’, just by virtue of their larger than average Size compared to average and late maturing peers. The relationship between Weight and height and adiposity is dependent upon stage of maturation and varies by race and gender (Eveluth & Tanner, 1990; Troiano & Flegal, 1998), and cannot always be col‘rected by knowing the sexual maturation stage of the youth (Hoerr et al., 1992). For these reasons, BMI alone is of questionable utility for clinical diagnosis during the most dynamic stage of adolescent development — ages 11-16 years. BMI can still be useful, 110vVever, to examine the proportion of the population who are overweight, as long as InVestigators recognize the indicator’s limitations for adolescents and use additional Infiasures of health risk. 29 ‘J In the present research Body Mass Index (BMI) was calculated to categorize a population of adolescents as overweight/obese, underweight or average weight to control for the effect of growth spurts and early pubescent fat. BMI is a convenient, public health indicator of weight trends, and because the CDC promotes the use of BMI growth charts to assess obesity, this study was consistent with what is already being used in society to assess weight category of adolescents. \ 3 Using this criteria and NHANES 111 data, 11% of 6-1 lyr children and 11% of youth agfid 12-17 yr are obese (Troiano and F legal, 1998). 30 METHODS Data source & participants The data used for this study were extrapolated from the National Longitudinal Study of Adolescent Health (Add Health), for which information was collected in all 50 states. A cluster sampling design was used to recruit subjects, and data were collected in a variety of ways and settings. The “In-Home” interview data, collected by interviewers in each adolescent’s home, were the focus of this study. One variable was used from the parent interviews to assess parent education. The parent interviews were conducted separately from the adolescents. When investigators use the appropriate weighting, this sample is nationally representative of all non-institutionalized adolescents in the United States. The large sample size (n~11,000) provides statistical power for many important comparisons among adolescents. Interview data were collected by the National Opinion Research Center of the University of Chicago in two waves: Wave I from September 1994 through December 1 995, and Wave II from April 1996 through August 1996. For the current cross-sectional Study, only Wave II of the data set was used, because Wave II had more detailed qUBStions on nutrition, and because heights and weights were measured by the Intel‘viewers, not self-reported as compared to data in Wave 1. The Wave II data was collected during the spring and summer months. 31 Sampling and weighting This research used a multistage sample selection in which both clustered selection and stratified selection procedures were applied, Figure 3.1‘. The database from Quality Figure 3.1 Sampling information from the Add Health data Students from Alternate Schools my; I: m Students from 132 Schools . Nationally Genetic Sample Genetic Sample Representative No Sample Sample Weights Weights Available " l I l. I lNDVGI In—Home N=l.82l Wave I ln-Home N218.924 Wave ll Iii-Home N=13,570 Education Data, Inc., (QED) was used as the sampling frame for Add Health. In this database a total of 26,666 public and private high schools were listed and was thought to be the most comprehensive list of high schools available in the US (Tourangeau & Hee- ChOon 1998). \ 4 Bearman, Peter S., Jones, Jo, and Udry, J. Richard. (1997) The National Longitudinal Study of Adolescent Health: Research Design [WWW document]. URL: L‘Ltg://www.m.unc.edu/Qroiects/addhealth/dengnhtml 32 A sample of 80 eligible high school clusters was selected by a systematic sampling method by using eight variables to sort the schools including size, type (public/ private/ parochial), urbanicity status, percent of white, percent of black, grade span, census region, and census divisions. All students in the selected school were asked to complete the In-School Questionnaire. Each participating school provided a roster of its students. For the stratified selection procedures, the adolescents for the In-Home interview were selected from the list of students who completed an In-school Questionnaire plus those who did not complete a questionnaire but were listed on a school roster. The students in the eligible list were stratified by grade and sex and were selected based on 12 sex-by-grade strata. In sum, a total of 20,745 adolescents in Grades 7 through 12 in the US were interviewed during Wave I (78.9% response rate), while a total of 14,738 adolescents Were interviewed during Wave II (88.2% response rate), which was limited to students from Wave I who were still in school. The multistage sample design serves as a basis for inferences from the Add Health sample to the adolescent population in the US (Frankel 1983) The multi-staged, cluster sample presents a complex experimental design; analyses must take this design into account. Estimates of variance, standard errors, and, theI‘efore, confidence intervals and inferential parametric tests will otherwise be incorrect. StEltistical packages such as SAS and SPSS warn not to use their package with such Satnpling designs, especially for variance estimates or comparisons. \ 50f the 80 selected schools, 52 were eligible and agreed to participate. The remaining 28 Schools were replaced by Similar high schools from the initial sample. 33 Because Add Health was a multiple level sampling design, the proper weight will differ depending on the sampling instruments used. The Add Health group has issued several documents about the weighting appropriate for various studies (Tourangeau & Hee- Choon 1998). Using the proper weights with the necessary Special software programs, such as SUDAAN will yield proper point estimates, such as means and regression parameters. SUDAAN is specifically designed for the analysis of cluster-correlated data from studies involving longitudinal data and multistage sample designs (SUDAAN website, http://www.rti.org/patents/sudaan/survey_research.html). This analysis was conducted with the Wave II data from In-Home interviews. Questions in Wave 11 included specific nutrition questions related to the topic of this research. The weighted data were used as appropriate in the all analyses (Bearman, Jones, Udry, 1997). The participants of Wave II were first separated by gender, and then, only those who reported starting puberty were used in the analyses. Next, each participant was categorized by the calculated Body Mass Index and whether they used weight control or not. Figure 3.2 shows the selection of variables. 34 Figure 3.2. Selection of subjects based on specific criteria in Wave [1 of Add Health (N=l 1,414) Females Males n=6963 n=6605 Menarche Sexual Maturation (yes) n=6063 Stage 3 (yes) n=4958 Underweight Average Underweight Average Overweight Overweight n=l 19 n= 3328 n=l479 l | 1 Weight Control Weight Control Weight Control Weight Control (Yes. n = 3219) (Yes, 11 = 1535) (Yes, 11 = 1566 ) (Yes, it = 1156) I l Descriptors: *Risky behavior (Yes/No) — suicide, substance use, tobacco use, drug use, dangerous weight control Types of weight control lpehaviors — diet, exercise, meal Skipping, laxative, vomiting, diet pills V Health-promoting behaviors V l l Nutrition Physical Activity I Fast Food Beverage *Responses to risky behaviors were Breakfast entered directly into a computer Food Groups whereas the rest of the variables were answered verbally to an interviewer. 35 Research instrument The In-home Interview data were recorded on portable computers during the interviews. This interview included questions on health status, health facility use, foods consumed, psychological well-being, peer networks, family composition and dynamics, educational aspirations and expectations, the ordering of events in the formation of romantic partnerships, sexual partnerships, decision-making processes, employment experience, substance use, puberty status and criminal activities, and the joint occurrence of risk behaviors. In Wave II, heights and weights were both self-reported by the adolescents and measured by the interviewers. Sensitive information, such as that on sexual behavior and risky health behaviors, was asked over headphones and adolescents entered responses directly into the portable computers for privacy. Definition of Variables Weight control Weight control, the outcome variable, is a dichotomous variable of “using weight control methods in the previous seven days” versus “not using any weight control methods”. This weight control variable was formed from five questions from the In- Home survey. The questions state “During the past seven days, did you do the following to try to lose weight or to keep from gaining weight? Dieting, exercising, made yourself vomit, took diet pills, took laxatives, other, none”. Those individuals who marked “yes” for any of the above weight loss methods were put into the weight control group and those who answered “none” were put into a non-weight control group. A sub-category of weight control use was created to reflect adolescents using potentially harmfiil weight control methods, which included vomiting, laxative use and 36 diet pill use to lose weight or to keep from gaining weight. This variable “dangerous weight control” was a dichotomous variable of “yes” or “no”. Weight control behaviors were reported by both adolescents “trying to lose weight” and by adolescents “trying to stay the same weight”. Therefore, adolescents trying to “lose weight” /“stay the same weight”, and reporting use of any weight control method were included in the “weight control use” variable. Healthy eating behaviors Healthy eating measures were dichotomous variables representing the health- promoting behaviors of eating and exercise. Food patterns selected to describe healthy eating included: 1) self-report of amount consumed from each of the five food groups during the previous day; 2) at least one glass of milk consumed during the previous day; 3) breakfast consumption or skipping; 4) self-report of the number of days fast food was consumed. Each of these food pattern questions was modified to represent a dichotomous outcome of healthy eating versus non-healthy eating behavior (Appendix A). The food pattern variables were used in bivariate analyses and in unconditional stepwise logistic regressions to determine the likeliness of healthy eating behaviors of youth using weight control. The first healthy eating behavior was defined as eating at least one serving from each of the following five food groups: fruit, vegetable, dairy, meat and grain (Schuette, Song, Hoerr, 1996). Table 3.1 shows the foods selected and categorization of each food group. 37 Table 3.] Selected foods categorized into five food groups from Add Health Wave [1 Fruit juice, apple, applesauce, pears, pineapple, bananas, grapes, berries, cherries, cantaloupes, melons, mangoes, papayas, oranges, grapefruit, Fru t tangermes, k1wrs, peaches, plums, nectarmes, apricots, rarsms, dried fruit Mixed vegetables, squash, green beans, peas, cabbage, broccoli, carrots, Vegetable field peas, chick peas, lima beans, kale, greens, lettuce, spinach, tomatoes, yam, potatoes, zucchini, vegetarian pizza Dairy Milk, yogurt, cottage cheese, cheese Meat Dried beans, lentils, soybeans, tofu, hot dogs, ground meat, steak, pork, meat pizza, chicken, tuna, other fish, eggs, ham, nuts, peanuts Grain Cereal, breakfast bars, bread, pasta, bagels, torillas, rice, crackers The second healthy eating behavior was the consumption of at least one glass of milk during the previous day. The consumption of milk daily is important, because it is very difficult for children to meet their calcium needs without a source of milk in the diet (Kennedy & Goldberg, 1995). The third food pattern variable was “breakfast eating” defined as eating a breakfast of three or more food groups at least five days a week. The definition of adequate breakfast of three or more food groups was taken from the USDA school breakfast program which requires servings from three or more food groups (Devaney & Stewart, 1998). The final variable describing a healthy eating behavior was the consumption of fast food less than two times a week. Research from 1998 Michaelson & Associates showed that adolescents aged 12-17 years ate fast food on average 2.13 times per week (USA Today, 1998), therefore this study focused on adolescents eating fast food less than the average times a week versus those eating it equal to or more than the average times per week. 38 Healthy exercise behaviors Healthy exercise was defined as participating in physical activity at least three times a week, but preferably all days of the week, for 30 or more minutes (USDA, 2000). Those adolescents who did not exercise at least three times a week for thirty minutes or more were categorized as having unhealthy exercise behaviors. The physical activity variable was created from the question “In the past week, how many times did you play an active sport or do exercise?”. There was no way to determine if this exercise was at a level that would promote unhealthy results instead of leading to health and balance. Risky Health Behaviors Participants entered their own risky behavior responses directly into a computer to ensure total privacy and to decrease bias in the response. The risky behavior variables selected here have been associated with weight control behavior in past research studies (Tomeo et al., 1999; Shisslak et al., 1998; Krowchuck, Kreiter, Woods, Sinal & DuPont, 1998; Neumark-Sztainer , Story , Dixon , Murray , 1998). Risky behaviors were any behavior placing the health of a youth at risk and included use of tobacco, alcohol and illicit drugs as well as thoughts of suicide. Tobacco use was determined by response to the question, “Do you smoke regularly? Yes or no”. Substance abuse was analyzed as a single dichotomous variable that combined all responses to illicit drug use. The questions were asked as follows: “Have you tried or do you use d_rgg_fl(? Yes or no”. The list of illicit substances included cocaine, inhalants, injected drugs, marijuana, LSD, PCP, ecstasy, ice, speed, heroin and others. 39 Alcohol use was asked as “Do you regularly consume alcohol?” Those adolescents using alcohol at any time (46%) were considered to be using risky health behaviors, and those not using alcohol at all were considered not to be partaking in risky behaviors. Sociodemographic Variables Sociodemographic variables used in the analyses included age, ethnicity and grade. Ethnicity categories included White, African American, Hispanic, Native American, Asian and “Other”. For this analysis, Native American, Asian and “Other” were combined as “Other”, because of the small number of responses for these categories. The mean age of pubescent adolescents was 16 years, standard deviation 1.6 and ranged from 11 to 23 years for girls and 11 to 21 for boys. Grade levels were categorized into four responses, 9‘h grade or below, 10‘h grade, 11“I grade, 12‘h grade or beyond. Age and grade of the adolescents were separated in the final analyses to identify differences in weight control use between these two categories. Parental variables Parental education was taken from a separate parental survey performed on the parents instead of the adolescents, because on the “In-Home” adolescent survey less than 10% of the adolescents actually reported their parent’s education. By this means, parental education was available for all of the adolescents in Add Health Wave II. This variable was use as an estimate of socioeconomic status of the family because family income was not reported in Add Health. This variable was only used to see if parental education was related to weight control use by adolescents. This was rank order variable 40 of number of years education of the mother or father. The variable did not specify which parent the information was reported by. Weight category variable from Body Mass Index Body mass index (BMI) (kg/m2) was calculated from measured heights and weights of the adolescents. Youth were measured in clothes without shoes. Weight was taken on a heavy-duty spring scale provided by the interviewer, and height was measured with a fabric tape measure. Interviewers were provided the following research protocol for anthropometric measurements: The procedures are as follows: 1) Before weighing and measuring the respondent have the adolescent remove his/her shoes, 2) Place scale on a hard flat surface and have the adolescent step on the scale. Record the respondent’s weight, 3) Using the tape measure record the respondents’, measuring him/her from toe to head. Have the adolescents place his/her toe on one end of the tape measure and then unroll the tape measure to measure to the top of his/her head., 4) Enter the respondent’s height and weight into the laptop (Goodman, Hinden & Khandelwal, 2000). Once BMIs were calculated for each adolescent for this analysis, the new 2000 pediatric grth charts from CDC were used to categorize the adolescents as under-, over- and average weight specific for gender and age (http://www.cdc.gov/nchs/about/major/nhanes/growthcharts/clinical_charts.htm). Underweight was defined as a calculated BMI less than or equal to the 5‘“ percentile, average weight was defined as a calculated BMI between the 6‘“ and 84‘“ percentiles, at risk for overweight was defined as a BMI greater than or equal to the 85‘“ percentile and less than the 95‘“ percentiles, and overweight is defined as greater than or equal to the 95‘“ percentile. The variable for BMI classification is called, “weight category”. A 41 crosstabulation of self-reported BMI versus measured BMI was run and confirmed that measured weights and heights were more accurate than self-reported. Table 3.2 and Table 3.3 Show the crosstabulation of self-report by measured for girls and boys. Table 3.2 A comparison of pubescent girl’s measured versus self-reported heights and weights used to calculate BMI from Wave II weLghted data (n=5788). x2=398.67 p<0.001 Measured BMI Underwt. Avergge Overwt. Obese Self- reported Underwt. 56% 43% <1 % 1% BMI Average 2% 93% 5% <1% Overweight 1 1% 71% 18% Obese <1 % 8% 90% Table 3.3 A comparison of pubescent boy’s measured versus self-reported heights and weights used to calculate BMI from Wave II weigmed data (nfl479). x2=110.22 p<0.001 Measured BMI Underwt. Average Overwt. Obese Self- reported Underwt. 47% 47% 6% BMI Average 1% I 94% 5% <1% Overweight 17% 47% 47% Obese 2% 1% 94% Those youth of average weight or obese were the most accurate in self-reported heights and weights, whereas those falling in the categories of underweight and overweight had the most difficulty with accuracy of self-reports when compared to measured heights and weights. 42 Puberty Variable Puberty status of girls was a dichotomous variable from the question, “Have you ever menstruated? Yes or No”. Those girls saying “yes” were considered “pubescent”. , Puberty status in boys was determined by looking at levels of development of secondary sex characteristics (Tanner, 1984). The variables coincide with the “sexual maturation stages” as show in Table 3.4. Sexual maturation of boys was a dichotomous variable with those boys in stage greater than three categorized as “pubescent” and less than three, “prepubescent boys”. Thus, only pubescent boys past their growth spurt were selected for analyses in this study. 43 Table 3.4 How puberty status was determined for boys in Add Health Wave II Sexual Maturation Stge‘I Stage] I Prepubescent Stage 2 I Pubic hair appears Genitalia growth Sweat gland action Pubic hair extends Genitalia growth Begin acne Voice changes Pubic hair thickens Genitalia growth Acne may be severe Voice deepens Stage 3 Stage 4 Stage 5 I Hair distribution increases I Genitalia fully mature I Acne may persist or increase aTanner, 1984 Question from Add Health No hair under arms at all. No hairs on face. Voice not lower. I have little facial and underarm hair. Yes, voice is a little lower than grade school I have some underarm hair, but not a lot. The hair is thick on my face. My voice is somewhat lower than grade school. I have lots of hair under my arms and it is thick. The hair is thick on my face like a grown man’s. Yes my voice is a lot lower than grade school. I have a whole lot of hair that is very thick, as much underarm hair as a grown man. The hair is very thick, like a grown man’s facial hair. My voice is a whole lot lower than when I was in grade school; it is as low as an adult man’s voice. Seventy-four percent of boys were at or passed stage three of the “Sexual Maturation Stages” in Wave II. Analyses Data Management The process of purchasing, obtaining and getting clearance to use the private-use data of Add Health was complex, frustrating and took six months. First the Institutional Review Boards([RB) approval was received on March 10, 2000. Then, the Carolina Population Center of the University of North Carolina Chapel Hill sent investigators a “Use of Sensitive Data Pack” for completion. Some of the requirements to obtain the data included: 1) principal investigator had a Ph.D. or other terminal degree and held a faculty appointment or research position at the receiving institution; 2) receiving institution was for higher education, research or a government agency and demonstrated record of using sensitive data according to commonly accepted standards of research ethics; 3) signed and submitted privacy agreement forms and security pledges; 4) a stand- alone password protected computer; 5)IRB approval of the stand-alone computer and of the project; 6)paper shredder on hand to destroy any unnecessary files relating to private use data; 7) list of public presentations at professional meetings using results based on these data and much more. IRB approval was received on March 10, 2000 and will last for one year. The entire process for obtaining the data took six months. Once the data were received, the files were unzipped and saved on SPSS 9.0 on a stand-alone password protected computer. The Wave H data set included 2540 variables from which the desired list of variables was selected. The final working database for this study was 270 original and created variables. After the working database was created, descriptive frequencies were run to determine outliers or unreasonable responses to questions. Unreasonable responses to drug questions (i.e. reporting using of any drug greater than 100 times in the last month) were labeled as missing data and not used in the analyses. A total of 30 respondents gave unreasonable responses. Respondents were removed from the database if their file lacked a weighting variable. The final database contained 11, 414 participants. Subjects included in the analyses were separated by gender and those who had indicated that they were pubescent as defined by the sexual maturation stages. The 45 separation of genders permitted for controlling differences in weight control use and the different reasons for weight control use by gender. Only pubescent adolescents were included in the study, because of the effect puberty has on weight and eating habits. Descriptive data were generated in SPSS 9.0 to describe the population characteristics. The statistical program SUDAAN (stand-alone version) was required to run statistical analyses to reflect the 2-stage cluster sampling design and to ensure p values did not get artificially deflated, as seen when using SPSS for a cluster sampling design. The statistically significant variables in the descriptive analyses (p<0.05) were used in the multivariate analyses of logistic regressions to report odds ratios and confidence intervals. Variables that were not significant, but have been significant in previous research findings, were still be included in the final model. Logistic regression was used for the multivariate analysis tool, because of its usefulness in predicting the presence or absence of an outcome based on values of a set of predictor variables (Norusis, 1999). The logistic regression analyses generated odds ratios which reported the magnitude of relationships between the outcome and the predictor variables while controlling for all other variables in the model. Reference values used in the logistic regression analyses are given an odds ratio of one, and each are designated in the results tables as such. The variances were estimated by a Taylor linearization method (SUDAAN User’s Manual, 1997). Confidence intervals are not significant when they include one. Table 3.5 shows detail of analyses run for each hypothesis. 46 Table 3.5 A description of analyses run for each hypothesis Hypothesis Analyses Hypothesis 1 There are differences in frequency of weight control use among pubescent youth by grade, race and age category. Hypothesis 2 Pubescent youth who use dangerous and non-dangerous weight control methods are at a higher risk to partake in risky health behaviors of tobacco use, drug use, alcohol consumption and suicide thoughts than those pubescent adolescents who do not use weight control. Hypothesis 3 There is a difference in risky health behaviors among pubescent youth using weight control classified as underweight, average weight and overweight/obese. Program: SUDAAN Analysis procedure: crosstabulation and logistic regression Weight control X age, gender, race, weight category Statistics reported: chi-square and significance (p<0.05) Results: variables with significant chi-square analyses controlled for in later logistic regressions. Program: SUDAAN Analysis procedure: crosstabulation after selecting for gender and puberty status Weight control X risky health behaviors Statistics reported: odds ratios, confidence intervals Program: SUDAAN Analysis procedure: crosstabulation after selecting for gender, puberty status and weight control use “yes” Statistics reported: odds ratios, confidence intervals Hypothesis 4 Youth of any weight category using weight control have unhealthy eating and exercise behaviors. Program: SUDAAN Analysis procedure: logistic regression after selecting for gender, puberty status and weight control use and including all significant findings from Ho 1-3 Statistics regorted: odds ratios, confidence intervals 47 RESULTS Eighty-three percent of boys from Wave II were determined to be at sexual maturation stage three or higher and 93% of girls from Wave II had started menstruation; only these pubescent adolescents were included in the analyses. The ages ranged from 11 to 18 years. Eighty percent of pubescent girls and 56% of pubescent boys used weight control methods at the time of the survey (Table 4.1). Weight control use existed in every weight category with even 31% of underweight girls and 30% of underweight boys using some form. Characteristics significantly related to weight control use for boys and girls included age, ethnicity, grade and weight category (x2, p<0.05) (Appendix C). Parental education was significant only for boys, with those practicing weight control more likely to have parents of a lower education level than those not practicing weight control. These categories were controlled for in later logistic regressions and further explained there. 48 Table 4.1 Demographic characteristics for pubescent girls and boys participating in Wave II of Add Health (N=11,4l4) . Girls‘ Boys‘I Demographic characteristics No % Yes, Wt. No % Yes, wt. ' Control‘ ' Control‘ Adolescent Total 6063 100 80 4958 100 56 Weight category *** Underweight" 161 3 31 119 2 30 Average weight‘1 4115 69 77 3328 68 46 Overweight‘ 922 15 91 722 15 70 Obese‘ 741 13 94 757 15 86 Age category *** 314 years 1120 18 82 769 16 63 15-16 years 2459 41 81 1941 39 56 17-18 years 2484 41 78 2248 45 51 Ethnicity ** ** White 3315 55 82 2887 58 57 African American 1379 23 78 857 17 46 American Indian 95 2 80 75 2 56 Asian 373 6 71 355 7 58 Hispanic 844 14 77 740 15 59 Other 48 1 74 40 l 57 Grade * 9th or younger 1903 34 81 1438 31 60 10th grade 1202 21 83 1000 22 57 11th grade 1313 23 78 1183 25 53 12th or beyond 1236 22 79 1010 22 52 Parental education * < High school diploma 932 15 77 636 13 56 _>_ High school diploma 2088 34 81 1676 34 57 303301163“ “011636 2310 38 80 2090 42 52 Parental nutrition control No 744 13 79 709 14 61 Yes 5139 87 81 4167 86 55 Weight control No 1261 2 0 2275 46 0 Yes 4799 79 100 2683 54 100 49 Table 4.1 (cont’). Fruit and Vegetable Intake Inadequate 3535 58 80 2732 55 56 Adequate (S-A-Day) 2522 42 80 2222 45 55 Milk Consumption No milk 2230 36.8 81 1172 24 56 At least 1 glass 3831 63.2 80 3786 76 56 Fast Food ** *** Too much 2043 44 77 1927 20 49 <2 days pg week 2606 56 82 1900 50 60 Breakfast Consumption * Skips breakfast 920 15 83 569 12 58 brkfst l-4x per wk 2425 40 82 1618 33 56 5-7xLer w 2718 45 78 2771 56 55 Exercise Behaviors ** ** Inadequate exercise 2913 48 78 2386 48 58 Adequate exercise 3150 52 82 2571 52 54 Smoking No smoking 4825 79.6 80 3917 79 57 Yes, smoking 1237 20.4 81 1038 21 53 Drug Use ** no drug use 4373 72 81 3489 70 58 yes, drug us 1690 28 77 1469 30 51 Suicide Thoughts No, suicide thoughts 5218 86 81 4537 92 56 Yes, suicide thoughts 834 14 81 416 8 56 Alcohol Consumption * No 3325 54.9 79 2632 53 58 Yes 2731 45.1 82 2318 47 53 'The variation in the number of subjects by characteristic is explained by missing data or refusal to respond. “Weighted percentage of weight control use = using dieting, exercise, laxatives, vomiting, diet pills to lose weight or to keep from gaining weight in the past seven days. “Underweight = from age and gender specific NCHS growth charts, < 5‘“ percentile dAverage weight = from age and gender specific NCHS growth charts, > 5‘“ percentile and <85‘“ percentile ‘Overweight = from age and gender specific NCHS growth charts, 2 85‘“ percentile and < 95‘“ percentile ‘Obese = from age and gender specific NCHS growth charts, 395‘“ percentile gW eight control reported in weighted percentages * p<0.05, ** p<0.01, *** p<0.001 To determine if adolescents reported weight control differently with another person present during the interview, a crosstabulation was run on weight control method 50 by whether a third person was present (Table 4.2). Of the total sample, 1916 (17%) respondents reported another person was there sometime during the interview. There was no significant difference for either boys or girls in reports of weight control use when another person was present. The response to “who was present” had various responses, but the most frequent choices included: mother (55%), sister (23%), brother (19%) and father (14%). Table 4.2 Crosstabulation of weight control uses by the variable “Was a third person present in the room during the interview?” o/o(no.) Was a third person present at time of the interview? BOYS No Yes no weight control 46% (1907) 43% (365) yes, wt control 54% (2206) 57% (475) (n=4953) 100% (4113) 100% (840) GIRLS N 0 Yes no weight control 21% (1055) 19% (204) yes, wt control 79% (3923) 81% (871) (n=6053) 100 % @978) 100% (1075) x2, boys (p=0.20) & girls (p=0.42) There were no significant differences in report of weight control use in youth either with or without a third person present in the room. However, when this analysis was run for type of weight control (Table 4.3 girls & 4.4boys), some significant differences were found. Girls were significantly more likely to report dieting and diet pill use, when a third person was not present in the room. 51 Tables 4.3 Crosstabulation of weight control methods of girls by the variable “Was a third person present in the room during the interview?” weighted % (no.) Was a third person present at time of the interview? GIRLS (n=4794) no yes Xz Significance Dieting No 71% (2789) 75% (655) 8.83 p<0.01 Yes 29% (1134) 25% (216) Exercise No 34% (1308) 39% (328) 3.50 us Yes 66% (2615) 61% (543) Vomiting N o 99% (4947) 99% (1072) 2.36 ns Yes 1% (33) 1% (4) Laxative use No 99% (3904) 99% (867) 0.00 us Yes 1% (l9) 1% (4) Diet pill use No 97% (3830) 99% (859) 4.21 p<0.05 Yes 3% (93) 1% (12) Tables 4.4 Crosstabulation of weight control methods of boys by the variable “Was a third person present in the room during the interview?” weighted % (no.) Was a third person present at time of the interview? BOYS (n=4958) No Yes Xz Significance Dieting No 87% (1929) 89% (423) 1.01 ns Yes 13% (277) 11% (52) Exercise No 37% (825) 38% (180) 1.27 us Yes 63% (1381) 62% (295) Vomiting No 99% (4106) 100% (840) 1.14 us Yes <1% (7) 0% Laxative use No 100% (2206) 99% (474) 1.00 nS Yes 0% <1% (1) Diet pill use No 99(2197) 99% (475) 5.59 p<0.05 Yes <1% (9) <1% 52 Chi-square analyses showed boys were significantly more likely to report diet pill use when a third person was not present in the room. Because the number of responses to diet pill use was a very small number (n~9), the chi-square results maybe artificially inflated, meaning that no conclusions could be drawn. A variety of weight control methods and combination of methods were used by both boys and girls. The types of weight control methods used are listed in Table 4.5. Any weight control methods reported by less than 1% of the adolescents are not listed as a percentage, but mentioned in the last row of the table. The majority of youth for all weight categories used exercise and/or dieting to control their weight. 53 Table 4.5. Weighted percentage of weight control methods used by girls and boys in Wave II of Add Health All weight categories Girls Boys Exercise 45 52 Dieting & exercise 18 9 Dieting 7 3 Other 0 2 Other &exercise 2 1 Dieting & other 1 0 Dieting exercise, & diet pill use 1 0 Underweight Exercise 11 11 Dieting & exercise 1 O Dieting 2 6 Other 0 7 diet, exercise, laxative use, & diet pill use 1 Average weight Exercise 48 46 Dieting & exercise 14 4 Dieting 5 1 Other 2 0 Other &exercise 2 0 Dieting & other 0 1 Dieting, exercise & diet pills I 0 Overweight/obese Exercise 42 60 Dieting & exercise 25 14 Dieting 9 4 Other 3 2 Other &exercise 2 1 Dieting, exercise & other 1 1 Dieting, exercise & diet pills I 0 Exercise & diet pills I 0 Other weight control methods reported by <1% girls: laxatives, vomiting, and any combination of diet, exercise, laxative, diet pills and vomiting Other weight control methods reported by <1% of boys: laxative use, diet pill use, vomiting and any combination of diet, exercise, vomiting, laxative use and diet pill use 54 More overweight/obese girls (25%) used diet and exercise than did average weight (14%) and underweight (18%) girls. Bivariate analyses showed weight category was significantly related to vomiting (p<0.05), diet pill use (p<0.01) and laxative use (p<0.05). Of girls who used dangerous weight control methods, there was a significant difference by weight categories for use. Those who were average weight were those most likely to use dieting, vomiting and diet pills with (x2=17.42, p<0.0001) 50% average weight, 30% obese, 20% overweight, <1% underweight. More overweight/obese boys (14%) used diet and exercise than average weight (4%) and underweight (0%) boys. Weight category was significantly related to diet pill use (p<0.05), exercise (p<0.01) and dieting (p<0.05), with overweight/obese, boys most likely to exhibit dangerous weight control behaviors compared to average weight or underweight boys. Hypothesis 1 There are significant differences in frequency of weight control use among youth by weight category, grade, race and age category. Although no significant differences were seen between grades and ages of girls for weight control use (Table 4.6), overweight/obese were three and five times more likely to use weight control than were average weight pubescent girls. Underweight girls were less likely to use weight control than were average weight girls. African American girls, Hispanics, and “Other” ethnic groups were less likely to use weight control than were White girls. Unlike girls, there were no differences in weight control use by ethnicity of boys. Older males were less likely to use weight control than the youngest group of boys ages 14 an below. Underweight, were less likely to use weight control, and overweight and obese boys were three and 11.0 times more likely to use weight control than average weight boys. 55 Table 4.6. Logistic regression analyses of demographic variables by weight control foigirls and boys particijating in Wave II of Add Health. OR’ 95% CI p Upper Lower ADOLESCENT GIRLS (n=4094) Weight Category Average weight 1.0 1.00 1.00 Underweight 0.1 0.04 .015 p<0.01 Overweight 3.6 2.50 5.23 p<0.01 Obese 5.6 3.45 9.18 p<0.01 Ethnicity White 1.0 1.00 1.00 African American 0.6 0.47 0.79 p<0.01 Hispanic 0.6 0.44 0.86 p<0.01 Other 0.7 0.48 0.98 p<0.05 Age category <=14 years 1.0 1.00 1.00 15-16 years 1.1 0.79 1.44 ns >=17 years 0.8 0.53 1.28 ns ADOLESCENT BOYS (n=3861) Weight control Average weight 1.0 1.00 1.00 Underweight 0.2 0.07 0.37 p<0.01 Overweight 3.9 2.96 5.02 p<0.01 Obese 11.5 8.54 15.60 p<0.01 Ethnicity White 1.0 1.00 1.00 Afiican American 0.7 0.45 1.10 ns Hispanic 1.2 0.81 1.81 us Other 1.4 0.91 2.14 ns Age category <=14 years 1.0 1.00 1.00 15-16 years 0.8 0.52 1.10 ns >=17 years 0.7 0.44 0.94 p<0.05 Parent education < High school diploma 1.0 1.00 1.00 High school diploma 1.2 0.66 2.17 us > High school diploma 1.5 0.78 2.69 ns OR = odds ratios aAn OR of 1 indicates the reference value. 56 Hypothesis 2 Pubescent youth who use weight control methods are at a higher risk for participating in risky health behaviors of tobacco use, drug use, alcohol consumption and suicide thoughts than those pubescent adolescents who do not use weight control participate. Hypothesis 3 There is a difference in risky health behaviors among pubescent youth using weight control classified as underweight, average weight and overweight/obese. No significant chi-square analyses relationships were found for girls between weight control use and risky health behaviors, however alcohol use (p<0.05) and drug use (p<0.01) in boys were significantly associated with weight control (Tables 4.7-4.8). A chi-square analysis between “dangerous weight control methods” by risky health behaviors in girls resulted in significant relationships with suicide (X2 = 7.28**), alcohol use (X2 = 13.95“”) and drug use (X2 = 14.79"). No significant associations were found between dangerous weight control and risky health behaviors in boys (data not shown). Data analyses to test Hypothesis 3 showed that: 1) drug use was significantly associated (p<0.05) to weight control use in underweight girls, 2) alcohol use was significantly associated (p<0.05) to weight control in average weight girls, 3) drug use was significantly associated (p<0.01) with weight control in average weight boys and 4) alcohol use was significantly associated (p<0.05) to weight control in underweight boys. (Tables 4.7 & 4.8). 57 Table 4.7 A comparison of risky health behaviors for girls who are using and who do not use weight control. Weight control All girls Underweight Average wt Overweight (n=6063) (n=161) (n=41 15) (n=l663) Yes Yes Yes Yes 4 Weighted % > Suicide thoughts X2 = 0.0 X2 = 0.09 X2 = 0.48 X2 = 3.40 No 85 86 87 82 Yes 15 14 13 18 Tobacco Use X2 = 0.9 X2 = 0.57 X2 = 0.92 X2 = 0.27 No 54 39 55 56 Yes 46 61 45 44 Drug Use X2 = 2.54 X2 = 368* X2 = 1.45 X2 = 0.33 No 71 82 71 73 Yes 29 18 29 27 Alcohol Use X2 =3.53 X2 = 0.28 X2 = 490* X2 = 0.32 No 84 72 83 87 Yes 16 28 17 13 *p<0.05, **p<0.01 Table 4.8 A comparison of risky health behaviors for boys who are using and who do not use weight control. Weight control All boys Underweight Average wt Overweight (n=495 8) (n=77) (n=3470) (n=1378) Yes Yes Yes Yes 4 Weighted % I Suicide thoughts X2 = 0.0 X2 = 2.73 X2 = 0.54 X2 = 0.54 No 90 75 91 90 Yes 10 25 9 10 Tobacco Use X2 = 0.0 X2 = 0.13 X2 = 0.20 X2 = 0.05 No 55 100 55 57 Yes 45 0 45 43 Drug Use X2 = 9.49** X2 = 1.51 X2 = 7.19** X2 = 1.49 No 69 45 68 69 Yes 31 55 32 31 Alcohol Use X2 = 5.26* X2 = 8.02** X2 = 2.55 X2 = 1.33 No 68 100 73 69 Yes 32 0 27 21 *p<0.05, **p<0.01 58 Hypothesis 4 Youth of any weight category using weight control have unhealthy eating and exercise behaviors. Logistic regression analyses were used to look at the relationships of weight control and health eating and exercise behaviors of girls and boys. Demographic variables significant in Hypotheses 1 were controlled for in this model including parental education. Also added, were risky health behaviors of smoking, alcohol, drug use and suicide, because these variables have been shown to be positively associated with weight control in previous research studies. 59 Table 4.9 Logistic Regression analyses of healthy behaviors of girls using weight control as reported in Add Health Wave II (n=3689) Healthy behaviors Weight control OR‘ 95% CI p Lower Upper Fruit and Vegetable Consump. Inadequate Intake 1.00 1.00 1.00 Adequate Intake 1.10 0.87 1.40 ns Milk consumption Inadequate milk 1.00 1.00 1.00 Adequate intake 0.99 0.79 1.24 ns Fast food consumption Too much fast food 1.00 1.00 1.00 Adequate fast food 1.54 1.23 1.92 <0.001 Breakfast consumption Breakfast 5-7 times/wk 1.00 1.00 1.00 Breakfast 1-4 times/wk 1.39 1.06 1.82 <0.05 Skipping breakfast 1.56 1.13 2.17 <0.01 Exercise behaviors Inadequate 1.00 l .00 1 .00 Adequate 1.59 1.26 2.02 <0.001 Race White 1.00 1.00 1.00 African American 0.54 0.40 0.75 <0.001 Hispanic 0.72 0.49 1.04 us Other 0.65 0.43 0.97 <0.05 Age <= 14 years 1.00 1.00 1.00 15-16 years 1.17 0.84 1.63 ns 17+ years 1.00 0.70 1.42 ns Weight Category Average weight 1.00 1.00 1.00 Underweight 0.10 0.06 0.20 <0.001 Overweight 3.65 2.51 5.30 <0.001 Obese 6.18 3.48 10.99 <0.001 Alcohol consumption No 1.00 1.00 1.00 Yes 1.58 1.23 2.05 <0.001 Drug Use No 1.00 1.00 1.00 Yes 0.60 0.45 0.81 <0.001 OR' — Odds ratios of 1.00 are not significant and represent the reference value. Parental education, smoking & suicide attempts not shown; results were insignificant. 60 Girls using weight control ate less fast food than those not using weight control (OR=1.54) (Table 4.9). Girls using weight control were also significantly more likely to skip breakfast (OR=1.56) and get adequate exercise (OR=1.59). In this model, girls using weight control were less likely to be Afiican American (OR=0.56), less likely to be underweight (OR=0.10) and more likely to have a BMI that put them into the category of overweight (OR=3.65) or obese (OR=6.18) than those not using weight control. Finally, girls using weight control were more likely to consume alcohol (OR=1.58), but less likely to use drugs (OR=0.60) than those girls not using weight control. No significant relationships were seen between smoking and weight control. Only one healthy eating and exercise behavior showed a significant relationship with weight control use in boys (Table 4.10). Pubescent/postpubertal boys using weight control methods ate less fast food than those not using weight control (OR=1.90). Overweight/obese boys were also more likely to use weight control (OR=6.15, 15.04) than normal weight boys. When controlling for all other variables in the final model on eating/exercise behaviors and weight control, parental education in boys was no longer significant. 61 Table 4.10 Logistic Regression analyses of healthy behaviors of boys using weight control as reported by Add Health Wave II. (n= 1299) Health behaviors OR' 95% CI p Lower Upper Fruit and Vegetable Consumption Inadequate Intake 1.00 1.00 1.00 Adequate Intake 1.03 0.74 1.43 ns Milk consumption Inadequate milk 1.00 1.00 1.00 Adequate intake 0.81 0.51 1.29 ns Fast food consumption Too much fast food 1.00 1.00 1.00 Adquate fast food 1.90 1.35 2.68 <0.001 Breakfast consumption Breakfast 5-7 times/wk 1.00 1.00 1.00 Breakfast 1-4 times/wk 1.20 0.82 1.76 ns Skipping breakfast 0.83 0.44 1.57 ns Exercise behaviors Inadequate 1 .00 1 .00 1.00 Adequate 0.83 0.59 1.18 ns Race White 1.00 1.00 1.00 Afiican American 0.53 0.30 0.92 <0.05 Hispanic 1.41 0.78 2.54 us Other 1.25 0.58 2.72 ns Age <= 14 years 1.00 1.00 1.00 15-16 years 0.38 0.23 0.63 <0.001 17+ years 0.43 0.21 0.92 <0.05 Weight Category Average weight 1.00 1.00 1.00 Underweight 0.09 0.01 0.67 <0.05 Overweight 6.15 3.88 9.76 <0.001 Obese 15.04 8.92 25.37 <0.001 Grade 9‘“ or below 1.00 1.00 1.00 10‘“ grade 1.97 1.09 3.55 <0.05 11‘“ grade 0.98 0.55 1.74 ns 12‘“ grade 1.44 0.55 1.74 ns Smoking No 1.00 1.00 1.00 Yes 0.82 0.55 1.23 ns OR" — Odds ratios of 1.00 are not significant and represent the reference value. 62 Further analysis of health behavior associations with weight control in various weight categories showed an interesting difference between average weight participants and overweight/ obese participants. A composite table to compare boys and girls of average weight and overweight/obese is shown in Table 4.11. No results could be reported for the underweight girls and boys using weight control and how weight control relates to their health behaviors, because the sample was too small. Table 4.11 Logistic regression analyses of use of health behaviors by average and overweight boys and girls. Average Weight Overweight Girls Boys Girls Boys Odds Ratio Comparisons Fast Food Consumption 1.55 fast food 2.05 fast food (Ref: Too much fast <2 /wk <2/ wk ns ns food) Breakfast 2.34 Brkfst, Consumption ns ns but inadequate ns (Ref: Adequate rkfst) amt Exercise behaviors 1 59adequate (Ref: Inadequate ' . ns ns ns . exercrse exercrse) Ethnicity 0.54 African 0.37 Afiican (Ref: White) American American “3 “5 Grade 1.7711‘“ 24910,), 0.1811‘“ 0.3811‘“ Qef: 9‘“) 2.62 12‘“ ' 0.11 12‘“ 0.23 12‘“ Age 0.31 15 (Ref: <=14 yr) 2.62 17+ yr tol6yr ns ns 0.22 17+yr 3:22:32) 1.56 yes ns 2.20 yes ns 31?:133; 0.63 yes ns 0.40 yes ns Ref. - reference value for logistic regression analyses. Variable set to 1.00 and not significant. ns=not significant 63 Average weight girls using weight control were significantly more likely to exercise adequately (OR=1.55), eat less fast food (OR=1.51) and drink alcohol (OR=1.56) and less likely to use drugs (OR=0.63) and be older (OR=1.77 to 2.62) than average weight girls not using weight control. Average weight girls using were control were 1.77 and 2.62 times more likely to be older, less likely to be African American or of “other” ethnic background (OR=0.54 & 0.58 respectively) than those average weight girls not using weight control. Results for overweight females differed somewhat from those for average weight females. Overweight females using weight control were still more likely to drink alcohol and were less likely to use drugs than those overweight females not using weight control. However overweight girls were less likely to be in older grades and 2.34 times more likely to eat some breakfast, even though it was still an inadequate amount. Average weight boys showed significant difference only for fast food consumption and weight control. Average weight boys using weight control were 2 times more likely to eat fast food less than twice a week than average weight boys not using weight control. Average weight, African American boys, and boys in the 10‘“ grade (OR=2.49) were more likely to be using weight control than 9‘“ grade and 11-12‘“ grade average weight boys. Average weight boys age 15-16 years were less likely to be using weight control (OR=0.22-0.31) than boys older or younger and average weight. None of the logistic regression analyses for overweight girls significantly predicted the health behaviors of those using weight control versus those not using weight control. Overweight/obese, boys using weight control were less likely to be in the 11‘“ or 12‘“ grade (OR=0.38 & 0.23) than overweight/obese boys not using weight control. Other 64 findings Show overweight/obese boys using weight control were less likely (OR=0.59) to get adequate exercise than overweight/obese boys not using weight control. Additional Statistics Weight control method by risky health behaviors Additional crosstabulation was run to look specifically at type of weight control use and how types of weight control related to risky health behaviors. The results for these analyses on girls and boys are in Table 4.12 and Table 4.13. Table 4.12 Are certain weight control methods associated with risky health behaviors in $15? (Weighted percentages) (N=6063) % Smoking DIES Alcohol Suicide No Yes No Yes No Yes No Yes Dieting X2 = 5.77* X2 = 450* No Yes No Yes No Yes No Yes No 52 48 71 29 82 18 56 14 Yes 54 46 71 29 84 16 83 17 Exercise X2 = 14.7** No 45 55 69 31 82 18 84 16 Yes 56 44 72 28 83 17 85 15 Vomiting No 52 48 71 29 83 17 85 15 Yes 46 54 37 63 67 32 71 29 Diet pills X2 = 9.98" X2 = 8.26M X2 = 496* No 53 47 72 28 83 17 85 15 Yes 45 55 45 55 49 21 75 25 Laxatives No 52 48 71 29 83 17 85 15 Yes 71 29 65 35 95 5 79 21 Other No 53 47 72 28 83 17 85 15 Yes 48 52 63 37 79 21 78 22 “Dieting for weight control” in girls was significantly associated with thoughts of suicide (p<0.05) and alcohol use (p<0.001). Other significant relationships for girls included “exercise for weight control” and smoking (p<0.01), “vomiting for weight 65 control” and drug use (p<0.01) or alcohol use (p<0.05), “diet pill use for weight control” and alcohol use (p<0.05) or drug use (p<0.01). Significant relationships for type of weight control use and risky health behaviors for boys were exercise and smoking (p<0.001). Table 4.13 shows weighted percentages describing the weight control methods by type of risky health behavior. Table 4.13 Are certain weight control methods associated with risky health behaviors in boys? (Weighted %) (N=4958) Smoking Drugs Alcohol Suicide No Yes No Yes No Yes No Yes Dieting X2 = 4.90* ns ns us No 50 51 69 32 69 31 91 9 Yes 59 41 64 36 59 41 86 14 Exercise X2 = 7.16** No 43 57 67 33 68 32 92 8 Yes 54 46 69 31 68 32 9O 10 Vomiting No 50 50 68 32 68 32 91 9 Yes 94 6 94 6 0 100 100 0 Diet pills No 50 50 68 32 68 32 91 9 Yes 70 30 18 82 40 60 100 0 Laxatives No 50 50 68 32 68 32 91 10 Yes 0 0 100 0 0 0 100 0 Other No 50 50 68 32 68 32 90 10 Yes 51 49 70 30 76 24 99 1 66 DISCUSSION Frequency of weight control The high prevalence of weight control use among youth and their associated health behaviors of eating and exercise were supported by the results of this study of the Add Health database. A high percentage of youth, 74% of girls and 44% of boys, were using some method of weight control. These percentages are the highest yet reported (Table 5.1). Karin et al. (1998) using the YRBS 1997 found slightly lower percentages of weight control in females (60%) and males (23%), and Story et al. (1998) using the Minnesota Adolescent Health Survey reported the lowest percentages of 27-43% of females and 8 -15% of males who were trying to control their weight. Table 5.1 A comparison of the prevalence of weight control by adolescents from three adolescent health databases. (Weighted percentages) Add Health Minnesota by YRBS Adolescent Chmielewski et by Health Survey al. Kann et al. by Story et al. Weight control use in girls 66 60 27-43 Weight control use in bgys 33 23 15 The Add Health results confirmed the findings from Kann et al. (1998) that girls report using weight control more than do boys. However, this study went further to show that average weight and underweight girls were also significantly more likely to use weight control than were average or underweight boys. As body weight increased in the Add Health participants, so did the percent of youth using weight control. These findings were supported by results from the MAHS, which also found overweight adolescents were more likely to practice weight control than their category of “non-overweight” youth (Story et al., 1995). 67 Ethnicity predicted differences among youth using weight control both in Add Health as well as in other studies (Herzog & Copeland, 1985). Results from the YRBS state that White students were significantly more likely than African American students to alter eating habits to lose weight (CDC, 2000). Story et al. (1998) reported that African American girls were less likely to diet in the past year than were White girls. Results of Add Health support Story’s conclusion that African Americans were significantly less likely to use weight control than were White girls (OR=0.61). In Add Health, parental education did not predict weight control use in girls, but did predict weight control use for boys. This contradicts findings from other studies in which the higher a parent’s education, the less likely adolescents were to use weight control. Parental education was reported by the parent in this study, removing the possibility of a misperception of parental educational level by the youth. This difference in reporting of parent’s education might account in part for differences in findings among surveys. In the Add Health “In-Home” youth interview, only 10% of adolescents reported knowing their parent’s education level. These findings from Add Health showed a decrease in weight control use as adolescent boys got older. Dieting and weight control tends to begin during adolescence (Richards et al., 1990), and as adolescents get older the drive for thinness decreases (Eislele, Hertsgaard & Light, 1986). Thirty-two percent of 9‘“ grade boys used weight control versus 24% of 12‘“ grade boys. Our findings were similar to those from the YRBS for which Story et al. (1998) reported that 30% of 9th graders and 20% of 12‘“ graders used some form of weight control. We found, however, no significant difference in prevalence of using weight control for girls by age group. Story et al. (1998) reported 68 on prepubertal as well as pubescent girls in the YRBS, whereas for this analysis of Add Health, we used only pubescent youth. Adolescence is a period of weight gain as both boys and girls progress through puberty (Allan, 1998). Girls gain a certain amount of fat in the early stages of puberty (Tanner, 1984), and if the entire population of girls have already started puberty and have experienced the “fat gain”, this may control for the aging as it relates to weight control. Pubescent girls may all use weight control methods at the same rate no matter their age, because they all have had a change in body composition due to puberty. The weight control use by boys by age group was irregular with the highest number of boys using weight control at ages 15-16 years (40%) followed by 17 years and older (39%), than 13-14 years (35%). This pattern of weight control differs from girls and might possibly reflect later physical and mental emotional and maturation of boys versus girls. The Youth Risk Behavior Survey did not show the same pattern of weight control increase in 15-16 year old boys, which may be associated with the effects sampling method. Weight control methods As in other studies, adolescents in Add Health reported using a variety of methods to lose weight or to keep from gaining weight such as: restrictive eating or dieting, exercise, vomiting, diuretics, laxatives and diet pills (CDC, 2000; Story, et al., 1998; Serdula, Collins, Williamson, Anda, Pamu, Byers, 1993; French, et al., 1995; Story, Rosenwinkel, Himes, Resnick, Harris & Blum, 1991; Button et al., 1997). Comparisons of weight control methods used by youth in this study to these from other large databases are shown in Table 5.2. The 64% of Add Health youth who reported using exercise for weight control was similar to that reported in YRBS 1999 (58%) 69 (CDC, 2000). We found slightly lower percentages of youth “dieting for weight control” (23%) compared to the 30% in YRBS, but higher than the 12% reported in the MAHS. Table 5.2 Comparison of weight control methods of youth from three different surveys. Add Health by Chmielewski et al. YRBS 1999 MAHS (weighted percentages) by CDC by Stoly et al.‘ All Girls Boys All Girls Boys N=7988 N=5099 N=2889 N=16,000 N=8207 N=7918 Dieting 23 28 13 30 Exercise 64 65 63 5 8 52 23 Diet pills 2 2 <1 6 Vomiting 1 1 <1 4 5 2 Laxative Use <1 <1 <1 ‘ Not nationally representative nr=survey did not report on specified behaviors The most striking difference in Add Health was that only 3% of girls and <1 % of boys reported use of dangerous weight control methods versus 10% in the YRBS and 7% in the MAHS. Differences in data collection methods among surveys may explain some of these variations in reports of dangerous weight control behaviors. Add Health questions could be asked of youth in the presence of an additional person, besides the adolescent and the interviewer, leading to some under-reporting of dangerous weight control behaviors. Because the Add Health survey included interviewer’s responses about who was present during the interview, we could determine that a third person present at any time during the interview did influence some weight control methods reported such as the use of diet pills and dieting. Add Health youth did under-report some behaviors when a third person was present. But less than 20% of the Add Health interviewers had a third person present during the interview and this additional person might not have been present during the weight control questions. On the other hand, the YRBS and MAHS 70 questions were asked in a school setting with other peers present. Peer influence likely could affected reporting and possibly led to some over-reporting of dangerous weight control behaviors in YRBS and MAHS. Thus, we conclude that weight control behaviors were slightly under-reported in Add Health and likely over-reported in YRBS and MAHS due to peer pressure. No published studies to date have examined how the weight category of adolescents using weight control relates to their eating and exercise behaviors. This analysis of Add Health demonstrates the importance of controlling for weight category when examining weight control use, because traits of average weight youth that predict weight control use differed from those of overweight youth. Average weight girls using weight control were more likely to exercise adequately, eat less fast food and drink alcohol and are less likely to use drugs than average weight girls not using weight control. Although overweight females using weight control were still more likely to drink alcohol and less likely to use drugs than those overweight females not using weight control, they were more likely to eat some breakfast, even though it was still inadequate. Average weight boys using weight control were two times more likely to eat fast food less than two times a week compared to average weight boys not using weight control. Also boys in the 10‘“ grade were 2.5 times more likely to be using weight control than 9‘“ grade and 11-12‘“ grade average weight boys. Weight control and risky behaviors Cigarette smoking in boys was significantly associated with “exercise for weight control” in Add Health which supports previous findings by Tomeo et a1. (1990). Unlike other studies, however, Add Health girls who used weight control were not more likely to 71 smoke (Tomeo et al., 1990; Story et al., 1998). Dieting for weight control in girls of Add Health was significantly associated with thoughts of suicide, as also found by Neumark- Sztainer in 1999 & Story et al. in 1994. Add Health also found that girls “dieting for weight control” were more likely to consume alcohol than were those not using weight control. Story et a1. (1994) also reported MAHS youth who drank were more likely to use weight control methods. Weight control and eating/exercise behaviors Twenty-eight percent of youth in Add Health reported eating the adequate amount of fruits and vegetables the previous day, just slightly higher than the 23% in YRBS. This Add Health survey was conducted spring and summer when fruits and vegetables should be readily available. This seasonal effect on the survey data should actually reduce the bias towards low intake of fruits and vegetables, because fruits and vegetables would have been less available in winter months. The 52% prevalence of adequate exercise used for weight control by Add Health participants was less than 65% reported in YRBS. In Add Health adolescents using weight control also reported an increase in physical activity (French et al., 1995). Story et al. (1999) showed that girls categorized as extreme dieters were less likely to eat fruits and vegetables than were moderate dieters or nondieters, and were more likely than more moderate dieters. Add Health researchers, however, found no significant relations for fruit and vegetable intake and weight control use. Story also reported moderate dieters ate two or more servings of fatty foods than nondieters. Add Health reported girls using weight control ate less fast food than those girls not using weight control, but we did not examine youth by category of weight control extremes as did Story. 72 Breakfast has been shown to be negatively affected by youth using weight control. Shisslak reported in 1996 that 14% of youth skipped breakfast for weight control, which was supported by the findings from Add Health. Strengths & limitations A strength of this study is that it is representative of non-institutionalized US adolescents enrolled in regular schools and these results can be applied to the US population. The new NCHS 2000 grth charts were also used in this study to correctly identify boys and girls as underweight, average weight and overweight. Weight control behaviors by the new weight categories is also described in the results of the study. Questions concerning weight loss behaviors and thoughts on body weight were asked verbally, which was a limitation of this study when another person was present during 20% of the ln-Home interviews. The use of dangerous weight control methods was likely reported at a lower frequency than actual occurrence. This bias was clearly demonstrated for use of weight control methods when crosstabulations were run on frequency of use by whether or not the interviewer recorded that a parent was present during the questioning. However, this limitation is in part a strength, because 80% of the interviews were private and not answered in classroom or gym settings as were the YRBS and MAHS. Another limitation of using the Add Health data is the lack of detail on some questions of interest or in how the questions were asked. Further analysis to determine the type of exercise and amount of exercise for weight control use is desired to clarify whether the high percent reported for “exercise for weight control” was a positive or negative health behavior. Weight control behaviors of parents has also been associated 73 with weight control behaviors in other family members, but no variables existed in Add Health to assess this relationship. Finally, food consumption was possibly over-reported because of the way each nutrition question was phrased, which is unlike the NHANES survey in which more standard nutrition assessment method was used, 24 hour recalls. 74 SUMMARY This study provides evidence that adolescents were in fact using weight control, and the methods of weight control were sometimes dangerous, especially when examined by weight category and gender. The high prevalence of use of weight control may be due to the increasing prevalence of obesity in all age groups, and the media portrayal of a thin ideal body image, but this cannot be confirmed by the Add Health survey. The results show that not just overweight, young white girls were using weight control methods, therefore, hypothesis one was rejected. Overweight adolescents reported the highest percentage of weight control use, but average weight and underweight youth also reported high percentages of use. White girls were more likely than any other ethnic group to use weight control, but Hispanic boys were more likely than White boys to use weight control. Average weight adolescents using weight control tended to be older and used the most dangerous methods, whereas overweight adolescents using weight control were younger students. Hypothesis two stated that youth using weight control were at a higher risk to participate in risky health behaviors of tobacco use, drug use, alcohol consumption and suicide thoughts. This hypothesis was not completely supported by the results of the study. Adolescents using weight control were more likely to consume alcohol, but less likely to use other drugs. Different types of weight control use were significantly associated with the different risky health behaviors. For girls, exercise was positively associated with tobacco use, diet pill use was positively associated with alcohol use and suicide thoughts and negatively associated with drug use. For boys, dieting and exercise were both positively associated with tobacco use. 75 Hypothesis three was to evaluate relationships between risky health behaviors and weight control use for adolescents of different weight categories. Average weight girls were more likely to consume alcohol when using weight control, and underweight girls were less likely to use drugs when using weight control. Average weight boys using weight control were less likely to use drugs, and underweight boys using weight control were more likely to consume alcohol. Finally, the last hypothesis questioned the eating and exercise behaviors of adolescents using weight control by looking at the consumption patterns of breakfast, fast food, fruits and vegetables, a variety of food, milk and the patterns of exercise. Girls using weight control were more likely to eat less fast food and get adequate exercise. They were also more likely to eat an inadequate breakfast or skip breakfast all together. No significant relationships were found between the rest of the eating behavior variables. The data show, however, that no matter if girls did or did not use weight control, only 28% of girls ate the minimally recommended five or more servings of fruits and vegetables a day, 14% had no servings of dairy the previous day and only 50% ate breakfast five or more times a week. These girls consumed a good variety of foods with 90% of the girls eating at least one serving from each of the five food groups the previous day. Boys were significantly more likely to eat less fast food when using weight control. No other eating or exercise variables were significant for boys. The data did show that 29% of the boys ate the recommended five servings of fi'uits and vegetables the previous day, regardless of weight control use. Nine percent of boys skipped dairy foods the previous day and 12% skipped breakfast. A high percentage of boys (93%) consumed a good variety of foods the previous day. 76 The above eating and exercise results were for a combination of boys and girls of all weight categories. When separating the youth by weight category, different eating and exercise behaviors were observed. Only average weight boys and girls were less likely to eat fast food. When separating average weight girls from the original model, no significant findings were demonstrated between weight control use and breakfast consumption, however average weight girls were the only group to report a significant relationship between weight control use and adequate exercise. Overweight girls were the only weight category to show a significant relationship between weight control use and breakfast consumption, although their breakfast consumption was still reported as inadequate. Weight control is not just seen in White, overweight girls, but was demonstrated by this study in every age group, ethnic group and by those in every weight category. Health and unhealthy eating and exercise behaviors differed among adolescents categorized as under-, average and overweight and using weight control. These results indicate that adolescents were not balancing eating and exercise behaviors to control weight as suggested by HP2010. 77 IMPLICATIONS The healthy guidelines state that a balance of eating and exercise will aid in decreasing the prevalence of obesity in adolescence. Adolescents were using weight control, and the weight control use was associated with both positive and negative eating behaviors, which do not represent the recommendations of the healthy guidelines. Adolescents should be encouraged to continue the reduction of eating fast food or selecting the healthy options that fast food restaurants offer, and keeping an increased amount of exercise, as long as it is not excessive. Adolescents also require finther counseling on the importance of fruit and vegetables consumption and on the effects of skipping breakfast and school performance. Health professionals should to be aware of the high percentage of underweight girls and boys using weight control. Not all adolescents will report their weight control behaviors in front of someone related to or close to them. This is important to know, because adolescents, especially those of average weight, were using dangerous methods to control their weight. Weight control use was also found to be positively associated with such risky health behaviors as alcohol consumption and suicide attempts and negatively associated with drug use. This study could not determine if the nondangerous weight control methods of diet and exercise were in fact “safe”, because of the lack of detailed items in Add Health. Another question on the extent of “dieting or exercise to control weight” might Show that the behaviors associated with risky health behaviors were in fact dangerous. 78 Even though obesity is increasing, the focus on obesity and the healthy guidelines in place do not seem to foster the correct health behaviors in adolescents. Adolescents using weight control may in fact be compromising their health status because of the eating behaviors associated with weight control use. 79 APPENDICES 80 APPENDIX A 81 QUESTIONS FROM ADD HEALTH USED TO CREATE VARIABLES Questions Parental education What was the last grade completed? thv—I “>09°>’.°‘ 5. 0. Eating behaviors Think about everything you had to each and drink yesterday. This includes snacks as well as your regular meals. 1. Did you drink ? 2. Yesterday did you eat. .. 3. In the last seven days, on how many days did you eat. .. a. at a fast food type place — McDonalds, Kentucky Fried Chicken, Pizza Hut, Taco Bell etc.? b. breakfast? 4. Do you currently take vitamins or minerals? Exercise behaviors 1. How many times did you exercise in the past week? Weight control behaviors 1. During the past seven days, which of the following things did you do in order to lost weight or to keep from gaining weight? a. dieted b. exercised c. made yourself vomit (1. took diet pills e. took laxatives f. other g. none Risky health behaviors Intro: Some teenagers have experimented with cigarette smoking, drinking alcohol and drug use. The questions that follow ask about your experience with these things. Remember, your answers will not be linked to you. 1. Do you smoke regularly? Yes or no 2. Have you tried or do you use drug X? Yes or no a. cocaine, inhalants, injected drugs, marijuana, LSD, PCP, ecstasy, ice, speed, heroin and others 3. How many times in the past year did you consume alcohol? 4. During the past 12 months, did you ever seriously think about committing suicide? 82 APPENDIX B 83 CHI-SQUARE ANALYSES OF DEMOGRAPHIC CHARACTERISTICS Table 4.1b Chi-square analyses of demographic characteristics for girls and boys participating in Wave II of Add Health (N=l3,570) Demtgraphic characteristics Chi-square P value* ADOLESCENT GIRLS Weight category 125.62 <0.001 Age category 3.56 0.17 Ethnicity 12.62 <0.01 Grade 4.50 0.22 Parent education 11.11 0.28 ADOLESCENT BOYS Weight category 228.47 <0.001 Age category 20.65 <0.001 Ethnicity 15.39 <0.01 Grade 9.75 <0.05 Parent education 16.73 <0.05 *Significant criteria x2 p<0.05 84 BIBLIOGRAPHY 85 10. ll. BIBLIOGRAPHY ADAF: American Dietetic Association Fact Sheet. Breakfast—Still kids most important meal of the day. [Online] (1996-1999). Available http://www.eatright.org/nfs/nfs$0.html, Accessed July 8, 1999. Allan, J. “New Directions for the Study of Overweight.” Western Journal of Nursing Research. 1998;20(1): 7-13. Barr S. Dieting attitudes and behavior in urban high school students: implications for calcium intake. J Adol Health. 1995;85:714-717. Beannan, Jones PS, Udry JR. "The National Longitudinal Study of Adolescent Health: Research Design" [Online] Available http://www.cpc.unc.edu/projects/addhealth/design.html, 1997, Accessed October 15, 2000. Blondell RD, Foster MB, Kamlesh DC. When your child is close to puberty. Am Fam Phys. 1999;60: 209-224. Bowen, DN, Tomoyasu, et al. The trip threat: A discussion of gender, class and race differences in weight. Women and Health. 1991;17:123-43. Brown KM, McMahon R, Biro FM, Crawford P, Schreiber GB, Similo SL, Waclawiw M, Striegel-Moore R. Changes in self-esteem in black and white girls between the ages of 9 and 14 years. J Adolesc Health. 1998;23:7-19. Bureau of Labor Statistics, US. Department of Labor. National Longitudinal Survey of Youth 1997 cohort, 1997-1999 (rounds 1-2) [computer file]. Produced by the National Opinion Research Center, the University of Chicago and distributed by the Center for Human Resource Research, The Ohio State University. Columbus, OH: 1999. Button EP, Loan P, Davies J, Sonuga-Barke EJ S. Self-esteem, eating problems and psychological well-being in a cohort of school girls aged 15-16: A questionnaire and interview study. Int J Eat Disord. 1997;21(1): 39-47. CDC. Youth Risk Behavior Surveillance — United States, 1999. Morbidity and Mortality Weekly Report. 2000;49(SSOS): 1-96. Center for Disease Control — Youth Risk Behavior Survey. How Are Adolescents Measuring Up Nutritionally? 1995 & 1997 Youth Risk Behavior Surveillance, American Medical Association. [Online] Available http://www.health.org/pubs/yrbbs/index.html, August 2000. 86 1..th ,n 1:1. 12. 13. 14. 15. l6. 17. 18. 19. 20. 21. 22. 23. 24. Connor-Greene, P. Gender differences in body weight perception and weight-loss strategies of college students. Women and health. 1988; 14:27-42. Cooper PJ, Goodyer I. Prevalence and Significance of weight and shape concerns in girls aged 11—16 years. British J Psychiatry. 1997;171:542-544. Crisp, A. Some possible approaches to prevention of eating and body weight/shape disorders with particular reference to anorexia nervosa Int J Eat Disord. 1988;7:1-17. Cusatis DC and Shannon BM. Influences on adolescent eating behavior. J Adolesc Health. 1996; 18(1):27-34. Daniels SR, Khoury PR, Morrison JA. The utility of body mass index as a measure of body fatness in children and adolescents: differences by race and gender. Pediatrics. 1997;99(6):804-807. Devaney B & Stuart B. 1998. Eating breakfast: Effects of the school breakfast program. US. Department of Agriculture, Food and Nutrition Service. Contract No. 53-3198-7-006. Dietz W, Tejedor E, Tejada J, Frank A. Neurologic complications casued by use of cocaine, amphetamines and symptoms. :Arch Neurobiol Madr. 1989;52(Suppl 1): 162-182. Dietz W. Use of body mass index (BMI) as a measure of overweight in children and adolescents. J Pediatrics. 1998;132: 191-3. Eisle J, Hertzgaard D, Light HK. Factors related to eating disorders in young adolescent girls. Adolescence. 1986;22(82): 283-290. Eveleth PB, Tanner JM. Worldwide variation in human growth. 2“‘1 ed. Cambridge, United Kingdom: Cambridge University Press, 1990. Feunekes GIJ, Braaf C, Meyboom S, Staveren W. Food choice and fat intake of adolescents and adults: Associations of intakes within social networks. Prev Med. 1998;27: 645-656. Field AE, Cheung L, Wolf AM, Herzog DB, Gortmaker SL, Colditz GA. Exposure to the mass media and weight concerns among girls. Pediatrics. 1999;103(3): 36-44. Frankel , M. (1983). Sampling theory. In P.H. Rossi, J .D. Wright, & A.B. Anderson (Eds), Handbook of survey research (pp. 21-67). San Diego, CA: Academic Press. 87 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. French SA and Jeffery R. Consequences of dieting to lose weight: effects on physical and mental health. Health Psych. 1994;13:195-212. French SA and Jeffery R. Weight concerns and smoking: a literature review. Ann Beh Med. 1995;17: 234-244. French SA, Perry CL, Leon GR, F ulkerson JA. Dieting behaviors and weight change history in female adolescents. Health Psych. 1995;14:545-555. Frisancho AR. Antrhopometric standards for the assenssment of grth and nutritional status. Ann Arbor, MI: University of Michigan Press; 1990. Garn S, LaVelle M, Piklington J. (1984). Obesity and living together. In DJ. Kallen & M. Sussman (Eds), Obesity and the family (pp 33-47). New York: Hayworth Press. Garn SM, Leonard WR, Hawthron VM. Three limitations of the body mass index. Am J Clin Nutr. 1986;44: 996-997. Gary, L., N. Brown, et al. (1989). Depression in Black American Adults: Finding from the Norfolk Area Health Study. Washington DC, Mental Health Research and Development Center, Institute for Urban Affairs and Research, Howard University. Gibbons K, Wertheim E, Paxton SJ, Petrovich J, Szmukler GI. Nutrient intake of adolescents and its relationship to desire for thinness, weight loss behaviors and bulimic tendencies. Aust J Nutr Diet. 1995;52:69-74. Goodman E, Hinden BR, Khandelwal 8. Accuracy of teen and parental reports of obesity and BMI. Pediatrics. 2000;106(1): 52-58. Gralen SJ, Levine MP, Smolak L, Mumen SK. Dieting and disordered eating during early and middle adolescence: Do influences remain the same? Int J Eat Disord. 1990;9(5): 501-512. Gritz ER and Crane LA. Use of diet pills and amphetamines to lose weight among smoking and nonsmoking high school seniors. Health Psych. 1991;10(5): 330-335. Harris E and Barraclough B. Excess mortality of mental disorder. Br J Psychiatry. 1998;173:1 1-53. Herzog D and Copeland P. Eating disorders. New Engl J Med. 1985;313(5):295. 88 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. Hoerr SL, Nelson RA, Lohman TR. Discrepancies among predictors of desirable weight for black and white obese adolescent girls. J Am Diet Assoc. 1992;92:450-453. Holdemess CC, Brooks-Gunn J, Warren MP. Co-morbidity of eating disorders and substance abuse: Review of the literature. Int J Eat Disord. 1994;16: 1-34. Huon GF and Strong KG. The Initiation and the Maintenance of Dieting: Structural Models for Large-Scale Longitudinal Investigations. Int J Eat Disord. 1996;23:361-369. Jeffery RW and French SA. Epidemic obesity in the United States: are fast foods and televison viewing contributing? Am J Pub Health. 1998;88(2): 277-280. Kann L, Kinchen SA, Williams BI, Ross JG, Lowry R, Hill CV, Grunbaum JA, Blumson PS, Collins JL, Kolbe LJ. Youth Risk Behavior Surveillance - United States, 1997. J Schol Health. 1997;68(9): 355-369. Keel PD, Lump KL, Leon GR, F ulkerson JA. (1998). Disordered eating in Adolescent Males form a School-Based Sample. Int J Eat Disord. 1998;23: 125- 132. Kennedy E, Goldberg J. What are American children eating? Implications for public policy. Nutrition Reviews. 1995;53: 111-126. Kilpatrick M, Ohannessian C, Bartholomew JB. Adolescent weight management and perceptions: An analysis of the National Longitudinal Study of Adolescent Health. J School Health. 1999;69(4): 148-152. King M. Eating disorders in general practice. BMJ. 1987;17:245-253. Kirkley B and Burge J. Dietary restrictions in young women: issues and concerns. Ann Beh Med. 1989;11: 66—71. Krebs NF, Johnson SL. Guidelines for health children: Promoting eating, moving and common sense. J Am Diet Assoc . 2000;100(1): 37-39. Krowchuck DP, Kreiter SR, Woods CR, Sinal SH, DuPont RH. Problem dieting behaviors among young adolescents. Arch Ped Med. 1998;152: 884-888. Kuczmarski RJ, Ogden CL, Grummer-Strawn LM, F legal KM, Guo SS, Wei R, Mei Z, Curtin LR, Roche AF Johnson CL. CDC growth charts: United States. Advance data from vital and health statistics; no. 314. Hyattsville, Maryland: National Center for Health Statistics. 2000. 89 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. Kuskowska-Wolk A, Karlsson P, Stolt M, Rossner S. The predictive validity of body mass index based on self-reported weight and height. Int J Obes. 1989;13: 441-453. Levine ML, Smolak L, Hayden H. The relation of sociocultural factors to eating attitudes and behaviors among middle school girls. J Early Adolesc. 1994;14(4): 471-490. Levine ML, Smolak L, Moodey AF, Shuman MD, Hessen LD. Normative developmental challenges and dieting and eating disturbances in middle school girls. Int J Eat Disord. 1994;15(1): 1 1-20. Lifshitz F, Moses N. A complication of dietary treatment of hypercholesterolemia. Amer J Dis Child. 1989;143:537-542. Lindeman AK. Quest for ideal weight: costs and consequences. Med Sciences in Sports & Exercise. 1998;31(8):1135-1140. Malina RM, Katzmarzyk PT. Validity of the body mass index as an indicator of the risk and presence of overweight in adolescents. Am J Clin Nutr. l999;70(suppl):]31S-136S. Morbidity and Mortality Weekly Report Surveillance Summaries. Youth Risk Behavior Surveillance: United States, 1999. [Online] Available http://www.cdc.gov/mmwr/preview/mmwrhtml/ss4905a1.htm, June 9, 2000. National Center for Health Statistics. Analytic and reporting guidelines. In: The Third National Health and Nutrition Examination Survey, NHANES 111 (1984—94). Hyattsville, MD: Westat Inc; 1996:1-47. NCHS Website. National Center for Health Statistics: Monitoring the Nation’s health. Clinical Growth Charts. [Online] Available http://www.cdc.gov/nchs/about/major/nhanes/growthcharts/clinical_charts.htm, October 16, 2000. Neumark-Sztainer DM, Story M, Dixon LB, Murray DM. Adolescents Engaging in Unhealthy Weight Control Behaviors: Are They at Risk for Other Health- Compromising Behaviors? Am J Pub Health. 1998;88(6): 952-955. Neumark-Sztainer D, Story M, Resnick MD, Blum RW. Lessons learned about adolescent nutrition fiom the Minnesota Adolescent Health Survey. J Am Diet Assoc 1998;98: 1449-1456. Neumark-Sztainer D, French SA, Jeffery RW. Dieting for weight loss: Association with nutrient intake among women. J Amer Diet Assoc. 1996;96:1172-1175. 9O 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. Norusis MJ. (1999). SPSS Regression Models 10.0: Logistic Regression. Pg.3. Ortega RM, Requejo AM, L0pe-Sobaler AM, Quintas ME, Andres P, Redondo MR, Navia B, Lopez-Bonilla MD, Rivas T. Difference in the breakfast habits of overweight/obese and normal weight schoolchildren. Int J Vit Nutr Res. 1998;68(2):125-132. Panel TAC. Methods for voluntary weight loss control: Technology Assessment Conferences Statement. Ann Int Med. 1992;116:942-949. Patton GC, Selzer R, Coffey C, Carlin JB, Wolfe R. Onset of adolescents eating disorders: Population based cohort study over 3 years. BMJ. 1999;318:765-768. Pesa JA, Turner LW. Ethnic and racial differences in weight-loss behaviors ll among female adolescents: Results from a national survey. Am J Health Studies. 1999;15(1):l4-21. Peters, P. K., R. J. Amos, et al. Questionable dieting behaviors are used by young adults regardless of sex or student status. J Am Diet Assoc. 1995;95:709-711. Psychiatry, American Academy of Child and Adolescent Psychiatry. Facts for families: Teenager with eating disorders. [Online] Available http:// www.aacap.org/factsfam/eating.htm, July 18, 1999. Pugliese, M., F. Lifshitz, Grad G, Marks-Katz M. Fear of obesity: a cause of short stature and delayed puberty. New England J Med. 1983;309: 513-518. Richards MH, Casper RC, Larson R. Weight and Eating Concerns Among Pre- and Young Adolescent Boys and Girls. Soc Adol Med. 1990,11: 203-209. Rogol AD, Clark PA, Roemmich JN. Growth and pubertal development in children and adolescents: effects of diet and physical activity. Am Society Clin Nutr. 2000;72(2): 521S-5288. Sardinha LB, Going SB, Teixeira PJ, Lohman TG. Receiver operating characteristics analysis of body mass index, triceps skinfold thickness, and arm girth for obesity screening in children and adolescents. Am J Clin Nutr. 1999; 70(6): 1090-1095. Schroeder DG & Martorell R. F atness and body mass index from birth to young adulthood in a rural guatemalan population. Am J Clin Nutr. l999;70(suppl): 1 37S-44S. 91 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. Schuette LK, Song WO, Hoerr SL. Quantitative use of the Food Guide Pyramid to evaluate dietary intake of college students. J Am Diet Assoc. 1996;96: 453- 457. Serdula MK, Collins E, Williamson DF, Anda RF, Pamu E, Byers TE. Weight control practices of US adolescents and adults. Ann Intern Med. 1993;119:667- 671. Shannon BM and Parks SC. Fast foods: a perspective on their nutritional impact. J Am Diet Assoc. l980;76(3): 242-247. Shisslak CM, Crago M, McKnight KM, Estes LS, Gray N, Pamaby OG. Potential risk factors associated with weight control behaviors in elementary and middle school girls. J Psychosomatic Res. l998;44(3/4):301-313. Stettler, N. Onset of adolescent eating disorders: Dieting may be an early sign, rather than a cause, of an eating disorder. BMJ. 1999;318: 1761. Story M, Neumark-Sztainer D, Sherwood N, Stang J, Murray D. Dieting status and its relationship to eating and physical activity behaviors in a representative sample of US adolescents. J Am Diet Assoc. l998;98(10): 1127-1135. Story M, French SA, Resnick MD, Blum RW. Ethnic/racial and socioeconomic differences in dieting behaviors and body image perceptions of adolescents. Int J Eat Disord. 1995; 18(2): 173-9. Story M, Rosenwinkcl K, Himes JH, Resnick M, Harris LJ, Blum RW. Demographic and risk factors associated with chronic dieting in adolescents. Am J Dis Child. 1999;145:994-998. Strauss, R. Self-reported Weight Status of Dieting in a Cross-sectional Sample of Young Adolescents: National Health and Nutrition Examination Survey III. Arch Ped Adol Mee. 1999;153(7): 741-747. SUDAAN User’s Manual, Release 7.5. Research Triangle Institute. Volume 11, page 10-2. SUDAAN website. SUDAAN and survey research. [Online] Available http://www.rti.org/patents/sudaan/survey_research.html, July 20, 1998. Swarr AE and Richards MH. Longitudinal effects of adolescent girls’ pubertal development, perceptions of pubertal timing and parental relations on eating problems. Develop Psych. l996;32(4): 636-646. 92 r. .. Emma‘s-J. I 87. 88. 89. 90. 91. 92. 93. Tanner, JM. (1984). Physical growth and developments. In J. O. F orfar & G. C. Ameil (Eds), Textbook of pediatrics (pp.278-330). Edinburgh, Scotland; Churchill Livingstone. Tomeo C, Field A, Berkey C, Colditz G, Frazier AL. Weight concerns, weight control behaviors and smoking initiation. Pediatrics. l999;104(4): 918-924. Tourangeau R, and Hee-Choon, S. National Longitudinal Study of Adolescent Health Grand Sample Weight. 1998. Chapel Hill, NC: Carolina Population Center, University of North Carolina at Chapel Hill. Troiano RP and F legal KM. Overweight children and adolescents: Description, epidemiology and demographics. Pediatrics. 1998;101: 497-504. USDA. United States Department of Agriculture and United States Department of Health an Human Services. Nutrition and Your Health: Dietary Guidelines for Americans. (2000). Home and Garden Bulletin No. 232. Fifth Edition. [Online] Available http://www.usda.gov/cnpp/DietGd.pdf, December 3, 2000. Wertheim EV, Mee V, Paxton SJ. Relationships among adolescents girls' eating behaviors and their parents' weight-related attitudes and behaviors. Sex Roles. 1999;41(4): 169-187. Yalcin SS and Kinik E. The impact of the sexual maturation stage on body mass index in adolescent girls. The Turkish J Ped. 1999;41(3):315-21. 93 11111 1111111 1111111 11111111111 3 1293 02094 815 -4