- Email: [email protected]
Couch potatoes or French fries: Are sedentary behaviors associated with body mass index, physical activity, and dietary behaviors among adolescents?
RESEARCH
Continuing Education Questionnaire, page 1433 Meets learning need codes 3020, 4040, 4060, and 4160
Couch potatoes or french fries: Are sedentary behaviors associated with body mass index, physical activity, and dietary behaviors among adolescents? JENNIFER UTTER, MPH, RD; DIANNE NEUMARK-SZTAINER, PhD, MPH, RD; ROBERT JEFFERY, PhD; MARY STORY, PhD, RD
ABSTRACT Objective To describe the demographic characteristics of adolescent boys and girls who engage in three sedentary behaviors (television/video use, computer use, and reading/ homework), and to explore how each sedentary activity is associated with body mass index (BMI), dietary behaviors, and leisure time physical activity. Design This study draws on data collected from Project EAT (Eating Among Teens), a school-based survey examining personal, behavioral, and socioenvironmental factors that are associated with nutritional intake among adolescents. Subjects The study sample consists of 4,746 middle and high school students from 31 public schools in a metropolitan area of the upper Midwest. All students were invited to participate. The overall response rate for Project EAT was 81.5%. Data collection was completed during the 1998-1999 school year. Statistical analyses Multivariate linear regression was used for examining associations between independent and dependent variables, controlling for age, race/ethnicity, and socioeconomic status. All differences were considered statistically significant at P⬍.05. Results Among boys, television/video use and time spent reading/doing homework were positively associated with BMI (P⬍.05), whereas for girls television/video and computer use were positively associated with BMI (P⬍.05). High television/ video use among boys and girls was associated with more unhealthful dietary behaviors (eg, increased consumption of soft drinks, fried foods, and snacks) (P⬍.05). In contrast, time spent reading/doing homework was associated with more healthful dietary behaviors (eg, increased consumption of fruits and vegetables) (P⬍.05). Leisure time physical activity was not associated with television/video use among boys or girls, but was positively associated with computer use and time spent reading/doing homework (P⬍.05). Applications/Conclusions Messages and advice aimed at reducing time spent in sedentary activities should be targeted at television/video use instead of time spent reading, doing homework, or using a computer. Nutrition education should incorporate messages about the influence of the media and advertising on dietary behaviors. J Am Diet Assoc. 2003;103: 1298-1305. 1298 / October 2003 Volume 103 Number 10
T
elevision has become the most ubiquitous medium in American households in the past three decades. The number of households with more than three television sets increased sevenfold from 1970 and 2000, and the number of households with cable television increased tenfold (1). Currently, 99% of adolescents have a television in their home, and 65% of adolescents have televisions in their bedrooms (2). Furthermore, approximately half of America’s adolescents report that the television is on “most of the time” at home (2). Because this changing media environment parallels a rapid increase in population obesity, especially among children and adolescents (3,4), time spent watching television has become a factor of interest in understanding the obesity epidemic. The association between time spent watching television and obesity has been noted previously (5-11). These findings are further supported by longitudinal studies showing increases in body mass associated with increased television viewing (5,6,9) as well as intervention studies demonstrating that decreasing time spent in sedentary activities, particularly television watching, is an effective component of obesity prevention programs for children and adolescents (12-15). To the best of our knowledge, previous studies have not explored the relationship between obesity and other sedentary activities, such as computer use and reading. Because previous interventions have either solely targeted television watching or targeted a group of sedentary behaviors, it is unknown which specific sedentary behaviors should be targeted for the most effective interventions. The impact of television viewing on obesity is likely due to both the displacement of time spent doing higher– energy expending activities and increased energy consumption (eg, more frequent snacking). Several studies have examined the
J. Utter, D. Neumark-Sztainer, R. Jeffery, and M. Story are with the Division of Epidemiology, School of Public Health, University of Minnesota, Minneapolis, MN. Address correspondence to: Dianne Neumark-Sztainer, PhD, MPH, RD, 1300 South 2nd St, Suite 300, Minneapolis, MN 55454. E-mail: [email protected] Copyright © 2003 by the American Dietetic Association. 0002-8223/03/10310-0001$30.00/0 doi: 10.1053/S0002-8223(03)01079-4
RESEARCH
association between television viewing and physical activity among adolescents (16-18) and have found that time spent in sedentary behaviors is inversely correlated with physical activity, particularly among girls. Other studies have lent support to the hypothesis that television viewing influences energy consumption. Results from the Third National Health and Nutrition Examination Survey show a positive association between time spent watching television and total energy intake among adolescents (8). Weekday television viewing has also been positively associated with eating at fast-food restaurants among adolescents (19). Few other studies have explored the relationships between television viewing and other behaviors with potential implications for obesity prevention in this population, though in a study of younger children, weekly television viewing hours correlated significantly with children’s energy intake and requests for foods commonly advertised (20). To our knowledge, no additional studies have examined the relationship of other sedentary behaviors, such as reading and computer use, to behaviors related to eating and body weight, such as consumption of fruits and vegetables, snacks, or soft drinks. The current study aimed to provide more information about specific sedentary activities (television/video use, computer use, reading/homework) and how they are associated with dietary behaviors, physical activity, and body mass index (BMI). In developing obesity prevention programs it is important to know if messages about reducing time spent in sedentary activities should apply to all sedentary activities or should target some activities instead of others. The research objectives for the current study are to: a) describe the demographic characteristics of adolescent boys and girls who engage in three different sedentary behaviors (television/video use, computer use, and reading/homework), and b) explore how each sedentary activity is associated with BMI, dietary behaviors, and physical activity. METHODS Study Population and Study Design This study uses data from Project EAT (Eating Among Teens) (19,21). The purpose of the larger study was to identify personal, behavioral, and socioenvironmental factors that are associated with nutritional intake among a population-based group of ethnically diverse adolescents. Data for Project EAT were collected during the 1998-1999 school year and included three phases: a) focus groups with adolescents, b) surveys and anthropometric measurements, and c) parent interviews. Twoweek test–retest reliability data for the student survey were collected from 167 adolescents from three schools (two urban junior-high schools and one suburban senior-high school). The present study used student survey and anthropometric data. The University of Minnesota Human Subjects Committee granted approval for the study. The study sample consisted of 4,746 middle- and high-school students, with 50.2% being male and 49.8% being female. The mean age of the sample was 14.9 years. Measurements were administered in 31 public schools in a metropolitan area of the upper Midwest, yielding a response rate of 81.5%. The main reasons for lack of participation were absenteeism and failure to return consent forms within schools requiring active consent. The overall study sample (n⫽4,746) was reduced for analyses to 4,480 due to students missing responses for the sedentary behavior variables. Missing responses among covariates in
multivariate analyses further reduced the effective sample size, but not by more than 20% of the reduced sample. Measures Sedentary behaviors Time spent participating in the three sedentary behaviors was assessed by the questions: “In your free time on an average weekday (Monday-Friday), how many hours do you spend. . . a. watching TV and videos, b. reading and doing homework, and c. using a computer (not for homework)?” and “On an average weekend day (Saturday or Sunday), how many hours do you spend. . . a. watching TV and videos, b. reading and doing homework, and c. using a computer (not for homework)?” Response categories for each of these questions were 0 hours through more than 5 hours. Test–retest correlations for weekday television/video (r⫽0.80), weekend television/video (r⫽0.69), weekday computer (r⫽0.66), weekend computer (r⫽0.71), weekday reading/homework (r⫽0.60), and weekend reading/homework (r⫽0.60) were all found to be acceptable. For each sedentary behavior (television/videos, using a computer, and reading/ homework), an hours-per-week variable was created by calculating a weighted sum of weekday and weekend use. The hoursper-day variable was created by dividing the hours-per-week variable by seven. Each of the three sedentary behaviors was also divided into a three-category variable: high use, average use, and low use. Cutpoints for the categories were made at the nearest whole hour-per-day at the 33rd and 66th percentiles of the distribution for each behavior. Therefore, equal numbers per category did not result and the cutpoints for each variable were not uniform. The television/video variable and the reading/homework variable were both nearly normally distributed. Because the computer use variable was highly skewed for boys and girls, the cutpoints for high, average, and low use are much lower than those for the other sedentary behaviors. Estimates for hours per day using a computer at the 25th percentiles for boys and girls were zero. At the 50th percentiles, they were 0.6 hours for boys and 0.5 hours for girls and at the 75th percentiles, they were 2 hours for boys and 1.4 hours for girls. For television/videos, high use is defined as 4 or more hours per day, average use is between 1 and 4 hours per day, and low use is 1 hour or less per day. For using a computer, high use is more than 2 hours per day, average use is 0.5 hour to 2 hours per day, and low use is less than 0.5 hour per day. For reading/homework, the high level includes more than 3 hours per day, the average level includes 1 to 3 hours per day, and the low level includes less than 1 hour per day. Dietary variables The dietary variables [energy; percent of energy from fat; dietary fat grams; and daily servings of soft drinks, fried food, snacks, fruits and vegetables (without french fries)] were assessed using the Youth Adolescent Questionnaire (YAQ), a semiquantitative food-frequency questionnaire. The daily servings of snacks variable was assessed with a cumulative measure of items 126 to 151, which ask about frequency of consuming specific snack foods. The YAQ is a 149item questionnaire (excluding demographic questions) that asks in more detail about foods commonly consumed by youth, while other less commonly consumed foods have been eliminated. This measure has been tested for reliability and validity on youth ages 9 to 18 years (22,23) and found to be within acceptable ranges for dietary assessment tools. To the best of Journal of THE AMERICAN DIETETIC ASSOCIATION / 1299
RESEARCH
Table 1 Mean number of hours per day spent in sedentary behaviors among boys and girls Television/Video
Boys Girls
n
MeanⴞSDb
2,240 2,240
2.80⫾1.45 2.55⫾1.49
P value
Computera MeanⴞSD 1.25⫾1.48 1.00⫾1.30
P⬍.001 a
Reading/Homework P value
MeanⴞSD
P value
1.92⫾1.40 2.35⫾1.44 P⬍.001
P⬍.001
Computer use does not include homework. SD⫽standard deviation.
b
our knowledge, the YAQ had not been tested for reliability and validity on an ethnically diverse sample such as the current study population. Before incorporating the YAQ into Project EAT, it was tested among a very–low income, ethnically diverse middle school population and it was found to be acceptable in terms of overall comprehension and ability to complete it within one class period. Trained research assistants administered the YAQ to students during school hours. Each YAQ was coded and checked for quality by a member of the research team who is a registered dietitian. The YAQ forms were then sent to Channing Laboratories (Boston, MA) for scanning and data report. Body mass index (BMI) Trained research staff measured height and weight using standardized equipment and procedures. The height measurement was taken with a portable stadiometer using the Frankfort Plane Technique. The weight measurement was assessed with a portable digital scale (SECA Integra 815; SECA Corporation, Columbia, MD) that was calibrated with a certified 50-pound weight. Staff were trained and tested by one member of the research team. Additionally, regular recertification and monthly quality-control checks were performed and found to be acceptable. BMI was calculated according to the formula: weight (kg) divided by height (m)2 (24). Physical activity variable Level of leisure time physical activity was assessed with a modified version of the Leisure Time Exercise Questionnaire (25). This measure has been shown to be reliable and to be significantly correlated with other measures of physical activity in children and adolescents (26). In this version, three questions were asked to assess the number of hours per week spent in strenuous-, moderate-, and mildintensity physical activity. Strenuous, moderate, and mild activities were assigned the metabolic equivalent (MET) values of 9, 5, and 3, respectively. MET values were used to approximate energy expenditure for the different levels of physical activity. Energy (kcal) expenditures were calculated as number of hours multiplied by the MET value for that activity and summed across the three levels of exercise intensity. Demographic variables Age was determined by self-report and divided into three age groups: age 13 years and younger, ages 14 to 16 years, and 17 years and older. Ethnicity/race was assessed using a single item of self-report of ethnic identity. Respondents were given the option of choosing multiple-responses; those reporting more than one response (other than white) were coded as “mixed/other.” Because only a small number of youth identified themselves as “Hawaiian or Pacific Islander,” these youth were coded as “mixed/other.” 1300 / October 2003 Volume 103 Number 10
The primary determinant of socioeconomic status (SES) was parental education level, defined as the higher level of educational attainment of either parent. Other variables considered in assessing family SES included family eligibility for public assistance, eligibility for free or reduced-cost school meals, and employment status of mother and father. An algorithm was developed to avoid classifying youth as high SES, based on parental education levels, if they were on public assistance, eligible for free/reduced school meals, or had two unemployed parents. These variables were also used to assess SES in cases for which there were missing data or “don’t know” responses for both parents’ education levels (n⫽1,058, 22.3%). Using the method of Classification of Regression Trees (27), these other variables were found to be predictive of parents’ education and reduced the number of missing SES values to 4.1% (n⫽196). SES levels as reported by adolescents were compared with SES levels as reported by a subsample of 861 parents of participating adolescents in Project EAT who were interviewed by telephone; correlations were found to be acceptable (r⫽0.68). Analysis The mean number of hours per day that boys and girls spent with each sedentary activity was stratified by the sociodemographic variables of interest (gender, ethnicity, SES, age) and compared using t tests (Tables 1, 2, and 3). The overall P values are reported for all of the variables. For race/ethnicity, SES, and age among boys and girls, post-hoc comparison tests were conducted to assess differences between pairs of groups; differences are considered significant at P⬍.05. Associations between sedentary behaviors and dietary, BMI, and physical activity variables were examined separately using multivariate linear regression with each sedentary behavior as the main effect. Age, race/ethnicity, and SES were adjusted for as covariates in these models. In a final step, the two sedentary behaviors not being tested as the main effect in the model were added to the models as covariates to predict each of the dietary, BMI, and physical activity variables. The sedentary behaviors were added as covariates to account for possible correlations between sedentary behaviors. All analyses were conducted stratified by gender using SAS statistical software (SAS version 8, Cary, North Carolina, 1999). RESULTS Demographic variables Mean number of hours per day participants were engaged in each sedentary activity is presented in Table 1. Boys and girls differed significantly in their sedentary behaviors. Boys spent significantly more time (2.8 hours per day) than girls (2.6 hours per day) with television/videos and computers, while girls
RESEARCH
Table 2 Sedentary behaviors (mean hours per day) by demographic characteristics in boys
Race/Ethnicity White Black Asian Hispanic Native American Mixed/Other P value SESc High Middle Low P value Age 13 years or younger 14–16 years 17 years and older P value
n
Television/Video MeanⴞSDb
Computera MeanⴞSD
Reading/Homework MeanⴞSD
1,185 349 404 137 69 71
2.78w⫾1.39 3.00x⫾1.49 2.86w⫾1.52 2.43y⫾1.42 2.61w,y⫾1.54 2.89w,x⫾1.47 P⫽.003
1.33w⫾1.50 1.04x⫾1.32 1.43w⫾1.54 0.82x⫾1.32 0.97x⫾1.28 1.31w,x⫾1.76 P⬍.001
1.93w⫾1.37 1.78w,x⫾1.33 2.36y⫾1.48 1.26z⫾1.23 1.55x,y⫾1.28 1.89w,x⫾1.61 P⬍.001
879 600 708
2.72⫾1.42 2.89⫾1.45 2.84⫾1.47 P⫽.058
1.35w⫾1.48 1.21w,x⫾1.45 1.18x⫾1.50 P⫽.042
2.13w⫾1.41 1.71x⫾1.31 1.84x⫾1.43 P⬍.001
619 1,326 295
2.90w⫾1.48 2.78w,x⫾1.42 2.64x⫾1.51 P⫽.032
1.30⫾1.46 1.23⫾1.47 1.25⫾1.59 P⫽.601
1.92⫾1.33 1.91⫾1.39 1.98⫾1.55 P⫽.697
a
Computer use does not include homework. SD⫽standard deviation. SES⫽socioeconomic status. w,x,y,z Means with different superscripts are significantly different (P⬍.05), means with the same superscripts are not. b c
Table 3 Sedentary behaviors (mean hours per day) by demographic characteristics in girls
Race/Ethnicity White Black Asian Hispanic Native American Mixed/Other P value SESc High Middle Low P value Age 13 years or younger 14-16 years 17 years and older P value
n
Television/Video MeanⴞSDb
Computera MeanⴞSD
Reading/Homework MeanⴞSD
1,041 406 460 110 89 100
2.35x⫾1.40 3.14y⫾1.51 2.45x⫾1.52 2.45x⫾1.35 2.66x,z⫾1.42 2.91z⫾1.45 P⬍.001
0.92z⫾1.21 1.06x,y⫾1.36 1.20x,y⫾1.43 0.82y,z⫾1.23 0.78y,z⫾1.24 1.03x,y,z⫾1.34 P⫽.001
2.41x⫾1.39 1.95y⫾1.38 2.89z⫾1.45 1.79y⫾1.30 1.87y⫾1.46 2.06y⫾1.26 P⬍.001
767 560 859
2.36x⫾1.45 2.67y⫾1.48 2.67y⫾1.47 P⬍.001
1.03⫾1.26 0.98⫾1.30 0.98⫾1.34 P⫽.688
2.61x⫾1.43 2.31y⫾1.38 2.15y⫾1.44 P⬍.001
652 1318 270
2.80x⫾1.52 2.48y⫾1.45 2.33y⫾1.46 P⬍.001
1.15x⫾1.37 0.94y⫾1.27 0.91y⫾1.26 P⫽.001
2.31⫾1.43 2.37⫾1.44 2.32⫾1.44 P⫽.715
a
Computer use does not include homework. SD⫽standard deviation. SES⫽socioeconomic status. x,y,z Means with different superscripts are significantly different (P⬍.05); means with the same superscripts are not. b c
Journal of THE AMERICAN DIETETIC ASSOCIATION / 1301
RESEARCH
Table 4 Dietary behaviors, BMI, and physical activity by each type of sedentary behavior among boys—adjusted for race, SES, age Computera
Television/Video High
Average
Low
High
Average
Reading/Homework Low
High
Average
Low
b
n⫽1279 n⫽344 n⫽533 n⫽807 n⫽900 n⫽478 n⫽1,053 n⫽709 n⫽617 4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™Mean⫾SEc™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™3 32.6⫾0.1 17.4⫾0.1 4.5⫾0.1 25.1⫾0.1 7.1⫾0.1 0.6⫾0.0 28.7⫾0.1 13.2⫾0.1 3.5⫾0.1
Total h/wk BMI (Body mass (index) 23.3x⫾0.2 Dietary behaviors Energy intake (kcal) 2,437x⫾55.0 Fat energy (% of total energy) 31.9x⫾0.0 Dietary fat (g) 41.1x⫾1.0 Soft drinks (daily servings) 1.60x⫾0.1 Fried food (daily servings) 0.66x⫾0.0 Snacks (daily servings) 3.60x⫾0.1 Fruit (daily servings) 2.14x⫾0.1 Vegetables (daily servings) 1.58x⫾0.1 Physical Activity Active Energy (kcal/kg per 66.1NS⫾1.5 week)e
23.2x,y⫾0.1
22.6y⫾0.3
2,178y⫾38.2 2,093y⫾71.2
23.1NS⫾0.2
23.0NS⫾0.2
23.2NS⫾0.2
2,350x⫾59.7
2,173y⫾48.4
2,223b⫾45.4
30.1y⫾0.0 35.4y⫾0.7
30.2y⫾0.0 32.6y⫾1.3
30.6NS⫾0.0 38.4x⫾1.1
30.5NS⫾0.0 35.6y⫾0.9
30.7NS⫾0.0 36.3x,y⫾0.8
1.39y⫾0.0
1.33y⫾0.1
1.52x⫾0.1
1.35y⫾0.0
1.47x⫾0.0
0.58y⫾0.0
0.55y⫾0.0
0.62NS⫾0.0
0.59NS⫾0.0
0.59NS⫾0.0
2.88y⫾0.1
3.06y⫾0.2
3.48x⫾0.1
2.87y⫾0.1
2.28x,y⫾0.1
2.47y⫾0.1
2.37NS⫾0.1
1.65x⫾0.1
1.81y⫾0.1
64.5NS⫾1.1
61.6NS⫾2.0
23.6x⫾0.2
23.0y⫾0.2
23.0y⫾0.2
2,231NS⫾63.3 2,227NS⫾42.8
2,248NS⫾1.0
29.9x⫾0.0 30.5x,y⫾0.0 35.5NS⫾1.2 36.5NS⫾0.8
31.2y⫾0.0 37.3NS⫾0.9
1.38x⫾0.1
1.40x⫾0.0
1.54y⫾0.0
0.57x⫾0.0 0.59x,y⫾0.0
0.63y⫾0.0
3.10y⫾0.1
2.97x⫾0.1
2.96x⫾0.1
3.41y⫾0.1
2.25NS⫾0.1
2.24NS⫾0.1
2.45x⫾0.1
2.30x⫾0.1
2.12y⫾0.1
1.76NS⫾0.1
1.63NS⫾0.1
1.62NS⫾0.1
1.92x⫾0.1
1.65y⫾0.1
1.49z⫾0.1
63.0x,y⫾1.7
67.0x⫾1.4
63.1y⫾1.3
69.5x⫾1.8
65.5x⫾1.2
59.7y⫾1.4
a
Computer use does not include homework. n reflects number of students in each level of sedentary behavior, not the final number for multivariate analyses. SE⫽standard error. d kcal⫽kilocalorie. e kcal/kg⫽kilocalories per kilogram of body weight. x,y,z Means with different superscripts are significantly different (P⬍.05); means with the same superscripts are not. NS not significant. b c
spent significantly more time than boys reading and doing homework. Mean numbers of hours per day spent with each sedentary activity by sociodemographic variables are displayed in Table 2 for boys and Table 3 for girls. Among boys, television use significantly differed by race/ethnicity and age, but not by SES. Black and Asian boys reported watching the most television, whereas Hispanic boys reported the least. Likewise, younger boys reported more time with television/videos than older boys. Among girls, television use was most common among black youth, girls in the middle- or low-SES groups, and girls age 13 and younger. Computer use was higher among Asian and white boys, and boys in the high- or middle-SES groups. There was no significant difference by age among boys reporting computer use. Asian or black girls reported the highest computer use along with girls aged 13 and younger. Unlike boys, among girls there was no significant association between computer use and SES. For both boys and girls, trends in time spent reading and doing homework were similar. Asian and white youth spent the most time in this sedentary behavior, as did youth reporting the highest SES group. Age was not significantly associated with time spent reading for either boys or girls. BMI and Sedentary Behaviors Associations between each of the sedentary behaviors and BMI, dietary behaviors, and physical activity are reported in 1302 / October 2003 Volume 103 Number 10
Table 4 for boys and Table 5 for girls. All associations have been adjusted for race/ethnicity, SES, and age. For both boys and girls, BMI was positively associated with television/video use. Boys reporting high television use had a mean BMI of 23.3, while boys reporting low television use had a mean BMI of 22.6. However, when computer use and reading/homework were added as covariates in the model, the association between television/video use and BMI was no longer significant (data not shown). Boys in the high reading group also had a significantly higher mean BMI than those reporting other levels of reading, but this association was no longer significant when the other sedentary activities were treated as covariates. Among girls, those reporting high television use and high computer use had mean BMIs (23.8) nearly one unit higher than those reporting low use of the respective activities. These findings remained significant when the other sedentary behaviors were treated as covariates. Dietary Intake and Sedentary Behaviors Dietary intake was significantly associated with most sedentary behaviors for both boys and girls. Energy intake for both boys and girls was positively associated with television/video and computer use. Boys reporting high television/video use consumed almost 400 kcal more per day than those in the low-use category. Similarly, girls reporting high computer use consumed more than 300 kcal more per day than girls reporting low use. Likewise, both boys and girls in the high television/
RESEARCH
Table 5 Dietary behaviors, BMI, and physical activity by each type of sedentary behavior among girls—adjusted for race, SES, age Computera
Television/Video Highb
Average
Low
High
Average
n⫽539
n⫽1,222
n⫽479
n⫽403
n⫽787
Reading/Homework Low
High
Average
Low
n⫽1,050
n⫽700
n⫽1,055
n⫽485
4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™Mean⫾SE ™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™3 32.2⫾0.1 16.8⫾0.1 4.4⫾0.0 23.9⫾0.1 7.0⫾0.1 0.5⫾0.0 28.8⫾0.1 14.0⫾0.1 3.8⫾0.0 c
Total h/wk BMI (Body mass index) Dietary Behaviors Energy intake (kcal)a Fat energy (% of total energy) Dietary fat (g) Soft drinks (daily servings) Fried food (daily servings) Snacks (daily servings) Fruit (daily servings) Vegetables (daily servings) Physical activity Active energy (kcal/kg per week)d
23.8x⫾0.2
23.3x,y⫾0.2
22.8y⫾0.2
24.1x⫾0.3
23.3y⫾0.2
23.1y⫾0.2
23.2NS⫾0.2
23.5NS⫾0.2
23.0NS⫾0.2
2,218x⫾59.0
1,936y⫾42.2
1,868y⫾36.2
1,991NS⫾45.3
1,924NS⫾36.6
1,963NS⫾53.9
2,118x⫾51.9
1,896y⫾33.8
1,918y⫾53.4
30.9x⫾0.0 33.3x⫾0.9
29.6y⫾0.0 28.1y⫾0.6
29.0y⫾0.0 27.9y⫾0.9
30.0NS⫾0.0 34.5x⫾1.0
29.8NS⫾0.0 29.3y⫾0.7
29.7NS⫾0.0 27.3z⫾0.6
29.0x⫾0.0 29.3NS⫾0.8
30.1y⫾0.0 29.0NS⫾0.6
30.2y⫾0.0 30.0NS⫾0.9
1.58x⫾0.1
1.27y⫾0.0
1.12z⫾0.1
1.51x⫾0.1
1.27y⫾0.0
1.27y⫾0.0
1.22x⫾0.0
1.28x⫾0.0
1.50y⫾0.1
0.64x⫾0.0
0.55y⫾0.0
0.53y⫾0.0
0.64x⫾0.0
0.54y⫾0.0
0.56y⫾0.0
0.54x⫾0.0
0.57x⫾0.0
0.62y⫾0.0
3.21x⫾0.1
2.78y⫾0.1
2.65y⫾0.1
3.43x⫾0.1
2.68y⫾0.1
2.77y⫾0.1
2.83x,y⫾0.1
2.74y⫾0.1
3.12x⫾0.1
2.24x⫾0.1
2.36x,y⫾0.1
2.57y⫾0.1
2.54x⫾0.1
2.43x,y⫾0.1
2.27y⫾0.1
2.54x⫾0.1
2.40x⫾0.1
2.09y⫾0.1
1.66x⫾0.1
1.77x⫾0.1
2.06y⫾0.1
1.98x⫾0.1
1.88x⫾0.1
1.69y⫾0.1
1.99x⫾0.1
1.78y⫾0.1
1.61y⫾0.1
45.9NS⫾1.6
48.2NS⫾1.0
49.4NS⫾1.6
52.5x⫾1.8
47.6y⫾1.3
46.4y⫾1.1
52.1x⫾1.4
47.3y⫾1.1
43.4z⫾1.6
a
Computer use does not include homework. n reflects number of students in each level of sedentary behavior, not the final number for multivariate analyses. c SE⫽standard error. d kcal⫽kilocalorie. e kcal/kg⫽kilocalories per kilogram of bodyweight. x,y,z Means with different superscripts are significantly different (P⬍.05); means with the same superscripts are not. NS not significant. b
video category were significantly more likely to consume more dietary fat and a higher percent of fat energy than the average and low users. In contrast, both boys and girls reporting high levels of time spent in reading/homework consumed significantly less percent of energy from fat than the low group. All associations remained significant when controlling for the other sedentary activities, with the exception of energy and computer use and dietary fat and computer use among boys. Both boys and girls who reported high use of television/videos consumed significantly more soft drinks, fried foods, and number of snacks per day than the average- and low-use groups. These findings contrast with findings for youth spending the most time with reading/homework. Both boys and girls who report high amounts of time spent with reading/homework consumed significantly fewer soft drinks, fried foods, and snacks per day than those in the low reading/homework category. Computer use was associated with more daily servings of soft drinks and snacks among boys and higher consumption of soft drinks, fried food, and snacks among girls. These findings all remained significant when adjusted for the other sedentary activities. High television/video use was associated with significantly
less consumption of fruits and vegetables among both boys and girls, whereas reading/homework was associated with higher consumption of fruits and vegetables. Computer use was also associated with higher consumption of fruits and vegetables for girls, but not for boys. These estimates changed only marginally and remained significant when the other sedentary behaviors were controlled for in the models. Physical Activity and Sedentary Behaviors Physical activity was not associated with television/video use for boys or girls, but was associated with computer use and time spent reading/doing homework. Boys in the average category of computer use and in the high category for reading/ homework expended more active energy than those in the low categories of the respective sedentary behavior. Among girls, those spending the most time with computers or reading/ homework reported significantly higher levels of physical activity than girls in the average and low categories for the respective sedentary activity. DISCUSSION The purpose of the current study was to describe the sociodemographic characteristics of adolescent boys and girls who Journal of THE AMERICAN DIETETIC ASSOCIATION / 1303
RESEARCH
spend time with various sedentary activities, and to describe how various levels of sedentary behaviors are associated with specific dietary behaviors, BMI, and physical activity. Boys and girls spent more time watching television/videos than with computers or reading/doing homework. On average, boys spend approximately 15 more minutes per day with television/videos than girls do. Likewise, adolescents age 13 and younger spend between 15 and 30 minutes more per day with television/videos than their older peers do. Both black boys and girls spent 3 or more hours per day with television/videos. Our findings are similar to other studies of media use among adolescents. A report by the Kaiser Family Foundation (2) also found that black youth spent more time with television than their white or Hispanic counterparts. Boys and girls spent on average twice as long with television/videos per day than with computers. Our findings are consistent with estimates of media availability in that 99% of all American teenagers live in a home with a television, whereas 73% of teenagers live in homes with a computer and less than 50% have Internet access (2).
Dietary intake was significantly associated with most sedentary behaviors for both boys and girls The most striking findings of the current research are the differences in the associations between three different sedentary activities and nutrient intake and foods consumption. High television/video use was associated with more unhealthful dietary behaviors among boys and girls, whereas those reporting inactivity due to reading/homework were more likely to report more healthful dietary behaviors. Differences in patterns of association may be due to differences in the sedentary behaviors. For example, while watching television, there are advertising influences and commercial breaks that may encourage snacking. In contrast, it would be necessary to take a break from reading/homework to snack. Alternatively, differences in patterns of association may be due to other environmental factors (eg, familial) influencing both the types of sedentary behavior children engage in as well as the types of food available at home. Differences in patterns of associations and potential explanations for these differences are discussed in more detail later in this section. Our finding that youth reporting high use of television/videos consume more energy per day is consistent with a similar crosssectional study of adolescents that found a positive association between time spent watching television and total energy intake (8). Likewise, our findings that television/video use was positively associated with BMI among boys and girls when controlling for race/ethnicity and SES are supported by several other studies (5-7). However, these previous studies did not control for other sedentary activities except videos and video viewing. In the current study, the association between BMI and television only remained significant for girls when the other sedentary activities were controlled for as covariates. This suggests that for girls, television is an independent correlate of BMI, whereas for boys there may be an association between time spent with television and other sedentary activities and BMI. 1304 / October 2003 Volume 103 Number 10
Among boys and girls, television/video use was positively associated with daily servings of soft drinks, fried foods, and snacks. Our finding may be in part explained by the role of advertising. Advertising that is targeted to children and adolescents includes mostly food and foods that are high in fat and/or sugar (28,29). Furthermore, food manufacturers allocate more than 75% of their advertising budget to television, and fast-food restaurants allocate more than 95% of their advertising budgets to television (30). Similarly, we found that computer use was positively associated with soft drinks and snack consumption among boys and girls. This may reflect the fact that food and beverage companies advertise their websites on television as well as use the Internet for drawings and sweepstakes. A study of the same population found that youth who frequented fast-food restaurants were more likely to watch television than youth who did not (19). This finding may explain the higher consumption of fried foods and soft drinks. It is of great interest that reading/doing homework was positively associated with many healthful dietary behaviors among boys and girls. To our knowledge, these associations have not been previously documented in the health and social sciences. It is also noteworthy that computer use was positively associated with fruit and vegetable consumption and physical activity among girls. Findings from a recent report showed that adolescents are likely to use the Internet for obtaining health information, particularly about chronic diseases and weight loss/ gain (31). Lastly, we did not find a significant relationship between television/video use and physical activity for either boys or girls. This may suggest that the role of television/videos in obesity may be through mediating consumption rather than energy expenditure. Other cross-sectional studies have demonstrated an inverse relationship between physical activity and television use (16-18). It is possible that we did not show similar findings due to the combination of television and video use in one variable and that our physical activity measure was self-report and based only on a three-item scale. Our study has several strengths, including a sociodemographically diverse sample of adolescents; inclusion of specific, validated measures of dietary intake; and assessment of BMI by trained research staff. Limitations include the combination of sedentary behaviors in questions and limited variety of sedentary activity measures. The study could be improved with questions asking about television and videos separately and by using more comprehensive measures of physical activity. Because our study design was cross-sectional, our findings cannot be used to determine causal pathways. Additionally, the current study only applies to youth attending school; it may be that current findings are accentuated among children who are not in school. Our findings may be further limited by other factors that may act as missing predictors (such as family and school environments) in the associations between the sedentary activities and dietary behaviors, BMI, and physical activity examined here. IMPLICATIONS The present study provides needed information for health professionals and researchers about adolescent health and nutrition. The findings suggest that messages and advice regarding reducing time spent in sedentary activities should be targeted at watching television/videos instead of time spent reading, doing homework, or using a computer. ■ Our findings also lend support to the hypothesis that the ■
RESEARCH
association between television viewing and obesity is moderated by increased consumption more than the displacement of time spent being physically active. Further research using stronger measures of physical activity may better support this hypothesis. ■ For dietitians, this study provides information about factors influencing adolescent eating behaviors and food choices. Nutrition education should incorporate messages about the influence of media and advertising on eating-related attitudes and behaviors. When providing nutrition advice about decreasing consumption of soft drinks, fast food, and snack foods, dietitians should consider the role of television in sending opposite messages. Advice about decreasing time spent being inactive should specifically target television viewing because it is so common among adolescents and is associated with poor dietary behaviors. References 1. 2000 Report on television. New York: Nielsen Media Research; 2000. 2. Roberts D, Foehr U, Rideout V, Brodie M. Kids & Media The New Millenium. Menlo Park, CA: The Henry J. Kaiser Foundation; 1999. 3. Troiano R, Kuczmarski R, Campbell S, Johnson C. Overweight prevalence trends for children and adolescents: The National Health and Nutrition Examination Surveys 1963 to 1991. Arch Pediatr Adolesc Med. 1995;149:10851091. 4. Prevalence of overweight among children and adolescents: United States, 1999. Washington, DC: National Center for Health Statistics; 2001. 5. Gortmaker S, Sobol A, Peterson K, Colditz G, Dietz W. Television viewing as a cause of increasing obesity among children in the United States, 19861990. Arch Pediatr Adolesc Med. 1996;150:356-362. 6. Dietz W, Gortmaker S. Do we fatten our children at the television set? Pediatrics. 1985;75:807-812. 7. Hernandez B, Gortmaker S, Colditz G, Peterson K, Laird N, Parra-Cabrera S. Association of obesity with physical activity, television programs and other forms of video viewing among children in Mexico City. Int J Obes Relat Metab Disord. 1999;23:845-854. 8. Crespo C, Smit E, Troiano R, Bartlett S, Macera C, Andersen R. Television watching, energy intake, and obesity in US children: Results from the Third National Health and Nutrition Examination Survey, 1988-1994. Arch Pediatr Adolesc Med. 2001;155:360-365. 9. Berkey C, Rockett H, Field A, Gillman M, Frazier A, Carmago C, Colditz G. Activity, dietary intake, and weight changes in a longitudinal study of preadolescent and adolescent boys and girls. Pediatrics. 2000;105:e56. 10. Dowda M, Ainsworth B, Addy C, Saunders R, Riner W. Environmental influences, physical activity, and weight status in 8- to 16-year olds. Arch Pediatr Adolesc Med. 2001;155:711-717. 11. Eisenmann J, Bartee R, Wang M. Physical Activity, TV Viewing, and Weight in U.S. Youth: 1999 Youth Risk Behavior Survey. Obes Res. 2002;10: 379-385. 12. Gortmaker S, Peterson K, Wiecha J, Sobol A, Dixit S, Fox M, Laird N. Reducing obesity via a school-based interdisciplinary intervention among youth: Planet Health. Arch Pediatr Adolesc Med. 1999;153:409-418. 13. Epstein L. Effects of decreasing sedentary behavior and increasing activity on weight change in obese children. Health Psychol. 1995;14:109-115.
14. Epstein L, Paluch R, Gordy C, Dorn J. Decreasing sedentary behaviors in treating pediatric obesity. Arch Pediatr Adolesc Med. 2000;154:220-225. 15. Robinson T. Reducing children’s television viewing to prevent obesity. JAMA. 1999;282:1561-1567. 16. Bungum T, Vincent M. Determinants of physical activity among female adolescents. Am J Prev Med. 1997;13:115-122. 17. Robinson T, Hammer L, Killen J, Kraemer H, Wilson D, Hayward C, Taylor C. Does television viewing increase obesity and reduce physical activity? Cross-sectional and longitudinal analyses among adolescent girls. Pediatrics. 1993;91:273-280. 18. Strauss R, Rodzilsky D, Burack G, Colin M. Psychosocial correlates of physical activity in healthy children. Arch Pediatr Adolesc Med. 2001;155:897902. 19. French S, Story M, Neumark-Sztainer D, Fulkerson J, Hannan P. Fast food restaurant use among adolescents: Associations with nutrient intake, food choices and behavioral and psychosocial variables. Int J Obes Relat Metab Disord. 2001;25:1823-1833. 20. Taras H, Sallis J, Patterson T, Nader P, Nelson J. Television’s influence on children’s diet and physical activity. J Dev Behav Pediatr. 1989;10:176-180. 21. Neumark-Sztainer D, Croll J, Story M, Hannan P, French S, Perry C. Ethnic/racial differences in weight-related concerns and behaviors among adolescent girls and boys: Findings from Project EAT. J Psychosom Res. 2002;53:963-974. 22. Rockett H, Wolf A, Colditz G. Development and reproducibility of a food frequency questionnaire to assess diets of older children and adolescents. J Am Diet Assoc. 1995;95:336-340. 23. Rockett H, Breitenbach M, Frazier A. Validation of a youth/adolescent food frequency questionnaire. Prev Med. 1997;26:808-816. 24. Diet and Health: Implications for reducing chronic disease risk. Washington, DC: National Research Council; 1989. 25. Godin G, Shepard R. A simple method to assess exercise behavior in the community. Can J Appl Sport Sci. 1985;10:141-146. 26. Sallis J, Buono M, Roby J, Micale F, Nelson J. Seven-day recall and other physical activity self-report in children and adolescents. Med Sci Sports Exerc. 1993;25:99-108. 27. Brieman L, Friedman J, Olshen R, Stone C. Classification of Regression Trees. Belmont, CA: Wadsworth International Group; 1984. 28. Kotz K, Story M. Food advertisements during children’s Saturday morning television programming: Are they consistent with dietary recommendations? J Am Diet Assoc. 1994;94:1296-1300. 29. Wilson N, Quigley R, Mansoor O. Food ads on TV: A health hazard for children? Aust N Z J Public Health. 1999;23:647-650. 30. Gallo A. Food advertising in the United States. America’s Eating Habits: Changes and Consequences. Washington, DC: USDA/ERS; 1999:173-180. 31. Rideout V. Generation Rx.com. Menlo Park, CA: The Henry J. Kaiser Foundation; 2001.
This study was supported by grant MCJ-270834 (D. Neumark-Sztainer, principal investigator) from the Maternal and Child Health Bureau (Title V, Social Security Act), Heath Resources and Service Administration, US Department of Health and Human Services. Preparation of this manuscript was supported by grant DK-50456 from the National Institute of Diabetes and Digestive and Kidney Diseases.
Journal of THE AMERICAN DIETETIC ASSOCIATION / 1305
Continuing Education Questionnaire, page 1433 Meets learning need codes 3020, 4040, 4060, and 4160
Couch potatoes or french fries: Are sedentary behaviors associated with body mass index, physical activity, and dietary behaviors among adolescents? JENNIFER UTTER, MPH, RD; DIANNE NEUMARK-SZTAINER, PhD, MPH, RD; ROBERT JEFFERY, PhD; MARY STORY, PhD, RD
ABSTRACT Objective To describe the demographic characteristics of adolescent boys and girls who engage in three sedentary behaviors (television/video use, computer use, and reading/ homework), and to explore how each sedentary activity is associated with body mass index (BMI), dietary behaviors, and leisure time physical activity. Design This study draws on data collected from Project EAT (Eating Among Teens), a school-based survey examining personal, behavioral, and socioenvironmental factors that are associated with nutritional intake among adolescents. Subjects The study sample consists of 4,746 middle and high school students from 31 public schools in a metropolitan area of the upper Midwest. All students were invited to participate. The overall response rate for Project EAT was 81.5%. Data collection was completed during the 1998-1999 school year. Statistical analyses Multivariate linear regression was used for examining associations between independent and dependent variables, controlling for age, race/ethnicity, and socioeconomic status. All differences were considered statistically significant at P⬍.05. Results Among boys, television/video use and time spent reading/doing homework were positively associated with BMI (P⬍.05), whereas for girls television/video and computer use were positively associated with BMI (P⬍.05). High television/ video use among boys and girls was associated with more unhealthful dietary behaviors (eg, increased consumption of soft drinks, fried foods, and snacks) (P⬍.05). In contrast, time spent reading/doing homework was associated with more healthful dietary behaviors (eg, increased consumption of fruits and vegetables) (P⬍.05). Leisure time physical activity was not associated with television/video use among boys or girls, but was positively associated with computer use and time spent reading/doing homework (P⬍.05). Applications/Conclusions Messages and advice aimed at reducing time spent in sedentary activities should be targeted at television/video use instead of time spent reading, doing homework, or using a computer. Nutrition education should incorporate messages about the influence of the media and advertising on dietary behaviors. J Am Diet Assoc. 2003;103: 1298-1305. 1298 / October 2003 Volume 103 Number 10
T
elevision has become the most ubiquitous medium in American households in the past three decades. The number of households with more than three television sets increased sevenfold from 1970 and 2000, and the number of households with cable television increased tenfold (1). Currently, 99% of adolescents have a television in their home, and 65% of adolescents have televisions in their bedrooms (2). Furthermore, approximately half of America’s adolescents report that the television is on “most of the time” at home (2). Because this changing media environment parallels a rapid increase in population obesity, especially among children and adolescents (3,4), time spent watching television has become a factor of interest in understanding the obesity epidemic. The association between time spent watching television and obesity has been noted previously (5-11). These findings are further supported by longitudinal studies showing increases in body mass associated with increased television viewing (5,6,9) as well as intervention studies demonstrating that decreasing time spent in sedentary activities, particularly television watching, is an effective component of obesity prevention programs for children and adolescents (12-15). To the best of our knowledge, previous studies have not explored the relationship between obesity and other sedentary activities, such as computer use and reading. Because previous interventions have either solely targeted television watching or targeted a group of sedentary behaviors, it is unknown which specific sedentary behaviors should be targeted for the most effective interventions. The impact of television viewing on obesity is likely due to both the displacement of time spent doing higher– energy expending activities and increased energy consumption (eg, more frequent snacking). Several studies have examined the
J. Utter, D. Neumark-Sztainer, R. Jeffery, and M. Story are with the Division of Epidemiology, School of Public Health, University of Minnesota, Minneapolis, MN. Address correspondence to: Dianne Neumark-Sztainer, PhD, MPH, RD, 1300 South 2nd St, Suite 300, Minneapolis, MN 55454. E-mail: [email protected] Copyright © 2003 by the American Dietetic Association. 0002-8223/03/10310-0001$30.00/0 doi: 10.1053/S0002-8223(03)01079-4
RESEARCH
association between television viewing and physical activity among adolescents (16-18) and have found that time spent in sedentary behaviors is inversely correlated with physical activity, particularly among girls. Other studies have lent support to the hypothesis that television viewing influences energy consumption. Results from the Third National Health and Nutrition Examination Survey show a positive association between time spent watching television and total energy intake among adolescents (8). Weekday television viewing has also been positively associated with eating at fast-food restaurants among adolescents (19). Few other studies have explored the relationships between television viewing and other behaviors with potential implications for obesity prevention in this population, though in a study of younger children, weekly television viewing hours correlated significantly with children’s energy intake and requests for foods commonly advertised (20). To our knowledge, no additional studies have examined the relationship of other sedentary behaviors, such as reading and computer use, to behaviors related to eating and body weight, such as consumption of fruits and vegetables, snacks, or soft drinks. The current study aimed to provide more information about specific sedentary activities (television/video use, computer use, reading/homework) and how they are associated with dietary behaviors, physical activity, and body mass index (BMI). In developing obesity prevention programs it is important to know if messages about reducing time spent in sedentary activities should apply to all sedentary activities or should target some activities instead of others. The research objectives for the current study are to: a) describe the demographic characteristics of adolescent boys and girls who engage in three different sedentary behaviors (television/video use, computer use, and reading/homework), and b) explore how each sedentary activity is associated with BMI, dietary behaviors, and physical activity. METHODS Study Population and Study Design This study uses data from Project EAT (Eating Among Teens) (19,21). The purpose of the larger study was to identify personal, behavioral, and socioenvironmental factors that are associated with nutritional intake among a population-based group of ethnically diverse adolescents. Data for Project EAT were collected during the 1998-1999 school year and included three phases: a) focus groups with adolescents, b) surveys and anthropometric measurements, and c) parent interviews. Twoweek test–retest reliability data for the student survey were collected from 167 adolescents from three schools (two urban junior-high schools and one suburban senior-high school). The present study used student survey and anthropometric data. The University of Minnesota Human Subjects Committee granted approval for the study. The study sample consisted of 4,746 middle- and high-school students, with 50.2% being male and 49.8% being female. The mean age of the sample was 14.9 years. Measurements were administered in 31 public schools in a metropolitan area of the upper Midwest, yielding a response rate of 81.5%. The main reasons for lack of participation were absenteeism and failure to return consent forms within schools requiring active consent. The overall study sample (n⫽4,746) was reduced for analyses to 4,480 due to students missing responses for the sedentary behavior variables. Missing responses among covariates in
multivariate analyses further reduced the effective sample size, but not by more than 20% of the reduced sample. Measures Sedentary behaviors Time spent participating in the three sedentary behaviors was assessed by the questions: “In your free time on an average weekday (Monday-Friday), how many hours do you spend. . . a. watching TV and videos, b. reading and doing homework, and c. using a computer (not for homework)?” and “On an average weekend day (Saturday or Sunday), how many hours do you spend. . . a. watching TV and videos, b. reading and doing homework, and c. using a computer (not for homework)?” Response categories for each of these questions were 0 hours through more than 5 hours. Test–retest correlations for weekday television/video (r⫽0.80), weekend television/video (r⫽0.69), weekday computer (r⫽0.66), weekend computer (r⫽0.71), weekday reading/homework (r⫽0.60), and weekend reading/homework (r⫽0.60) were all found to be acceptable. For each sedentary behavior (television/videos, using a computer, and reading/ homework), an hours-per-week variable was created by calculating a weighted sum of weekday and weekend use. The hoursper-day variable was created by dividing the hours-per-week variable by seven. Each of the three sedentary behaviors was also divided into a three-category variable: high use, average use, and low use. Cutpoints for the categories were made at the nearest whole hour-per-day at the 33rd and 66th percentiles of the distribution for each behavior. Therefore, equal numbers per category did not result and the cutpoints for each variable were not uniform. The television/video variable and the reading/homework variable were both nearly normally distributed. Because the computer use variable was highly skewed for boys and girls, the cutpoints for high, average, and low use are much lower than those for the other sedentary behaviors. Estimates for hours per day using a computer at the 25th percentiles for boys and girls were zero. At the 50th percentiles, they were 0.6 hours for boys and 0.5 hours for girls and at the 75th percentiles, they were 2 hours for boys and 1.4 hours for girls. For television/videos, high use is defined as 4 or more hours per day, average use is between 1 and 4 hours per day, and low use is 1 hour or less per day. For using a computer, high use is more than 2 hours per day, average use is 0.5 hour to 2 hours per day, and low use is less than 0.5 hour per day. For reading/homework, the high level includes more than 3 hours per day, the average level includes 1 to 3 hours per day, and the low level includes less than 1 hour per day. Dietary variables The dietary variables [energy; percent of energy from fat; dietary fat grams; and daily servings of soft drinks, fried food, snacks, fruits and vegetables (without french fries)] were assessed using the Youth Adolescent Questionnaire (YAQ), a semiquantitative food-frequency questionnaire. The daily servings of snacks variable was assessed with a cumulative measure of items 126 to 151, which ask about frequency of consuming specific snack foods. The YAQ is a 149item questionnaire (excluding demographic questions) that asks in more detail about foods commonly consumed by youth, while other less commonly consumed foods have been eliminated. This measure has been tested for reliability and validity on youth ages 9 to 18 years (22,23) and found to be within acceptable ranges for dietary assessment tools. To the best of Journal of THE AMERICAN DIETETIC ASSOCIATION / 1299
RESEARCH
Table 1 Mean number of hours per day spent in sedentary behaviors among boys and girls Television/Video
Boys Girls
n
MeanⴞSDb
2,240 2,240
2.80⫾1.45 2.55⫾1.49
P value
Computera MeanⴞSD 1.25⫾1.48 1.00⫾1.30
P⬍.001 a
Reading/Homework P value
MeanⴞSD
P value
1.92⫾1.40 2.35⫾1.44 P⬍.001
P⬍.001
Computer use does not include homework. SD⫽standard deviation.
b
our knowledge, the YAQ had not been tested for reliability and validity on an ethnically diverse sample such as the current study population. Before incorporating the YAQ into Project EAT, it was tested among a very–low income, ethnically diverse middle school population and it was found to be acceptable in terms of overall comprehension and ability to complete it within one class period. Trained research assistants administered the YAQ to students during school hours. Each YAQ was coded and checked for quality by a member of the research team who is a registered dietitian. The YAQ forms were then sent to Channing Laboratories (Boston, MA) for scanning and data report. Body mass index (BMI) Trained research staff measured height and weight using standardized equipment and procedures. The height measurement was taken with a portable stadiometer using the Frankfort Plane Technique. The weight measurement was assessed with a portable digital scale (SECA Integra 815; SECA Corporation, Columbia, MD) that was calibrated with a certified 50-pound weight. Staff were trained and tested by one member of the research team. Additionally, regular recertification and monthly quality-control checks were performed and found to be acceptable. BMI was calculated according to the formula: weight (kg) divided by height (m)2 (24). Physical activity variable Level of leisure time physical activity was assessed with a modified version of the Leisure Time Exercise Questionnaire (25). This measure has been shown to be reliable and to be significantly correlated with other measures of physical activity in children and adolescents (26). In this version, three questions were asked to assess the number of hours per week spent in strenuous-, moderate-, and mildintensity physical activity. Strenuous, moderate, and mild activities were assigned the metabolic equivalent (MET) values of 9, 5, and 3, respectively. MET values were used to approximate energy expenditure for the different levels of physical activity. Energy (kcal) expenditures were calculated as number of hours multiplied by the MET value for that activity and summed across the three levels of exercise intensity. Demographic variables Age was determined by self-report and divided into three age groups: age 13 years and younger, ages 14 to 16 years, and 17 years and older. Ethnicity/race was assessed using a single item of self-report of ethnic identity. Respondents were given the option of choosing multiple-responses; those reporting more than one response (other than white) were coded as “mixed/other.” Because only a small number of youth identified themselves as “Hawaiian or Pacific Islander,” these youth were coded as “mixed/other.” 1300 / October 2003 Volume 103 Number 10
The primary determinant of socioeconomic status (SES) was parental education level, defined as the higher level of educational attainment of either parent. Other variables considered in assessing family SES included family eligibility for public assistance, eligibility for free or reduced-cost school meals, and employment status of mother and father. An algorithm was developed to avoid classifying youth as high SES, based on parental education levels, if they were on public assistance, eligible for free/reduced school meals, or had two unemployed parents. These variables were also used to assess SES in cases for which there were missing data or “don’t know” responses for both parents’ education levels (n⫽1,058, 22.3%). Using the method of Classification of Regression Trees (27), these other variables were found to be predictive of parents’ education and reduced the number of missing SES values to 4.1% (n⫽196). SES levels as reported by adolescents were compared with SES levels as reported by a subsample of 861 parents of participating adolescents in Project EAT who were interviewed by telephone; correlations were found to be acceptable (r⫽0.68). Analysis The mean number of hours per day that boys and girls spent with each sedentary activity was stratified by the sociodemographic variables of interest (gender, ethnicity, SES, age) and compared using t tests (Tables 1, 2, and 3). The overall P values are reported for all of the variables. For race/ethnicity, SES, and age among boys and girls, post-hoc comparison tests were conducted to assess differences between pairs of groups; differences are considered significant at P⬍.05. Associations between sedentary behaviors and dietary, BMI, and physical activity variables were examined separately using multivariate linear regression with each sedentary behavior as the main effect. Age, race/ethnicity, and SES were adjusted for as covariates in these models. In a final step, the two sedentary behaviors not being tested as the main effect in the model were added to the models as covariates to predict each of the dietary, BMI, and physical activity variables. The sedentary behaviors were added as covariates to account for possible correlations between sedentary behaviors. All analyses were conducted stratified by gender using SAS statistical software (SAS version 8, Cary, North Carolina, 1999). RESULTS Demographic variables Mean number of hours per day participants were engaged in each sedentary activity is presented in Table 1. Boys and girls differed significantly in their sedentary behaviors. Boys spent significantly more time (2.8 hours per day) than girls (2.6 hours per day) with television/videos and computers, while girls
RESEARCH
Table 2 Sedentary behaviors (mean hours per day) by demographic characteristics in boys
Race/Ethnicity White Black Asian Hispanic Native American Mixed/Other P value SESc High Middle Low P value Age 13 years or younger 14–16 years 17 years and older P value
n
Television/Video MeanⴞSDb
Computera MeanⴞSD
Reading/Homework MeanⴞSD
1,185 349 404 137 69 71
2.78w⫾1.39 3.00x⫾1.49 2.86w⫾1.52 2.43y⫾1.42 2.61w,y⫾1.54 2.89w,x⫾1.47 P⫽.003
1.33w⫾1.50 1.04x⫾1.32 1.43w⫾1.54 0.82x⫾1.32 0.97x⫾1.28 1.31w,x⫾1.76 P⬍.001
1.93w⫾1.37 1.78w,x⫾1.33 2.36y⫾1.48 1.26z⫾1.23 1.55x,y⫾1.28 1.89w,x⫾1.61 P⬍.001
879 600 708
2.72⫾1.42 2.89⫾1.45 2.84⫾1.47 P⫽.058
1.35w⫾1.48 1.21w,x⫾1.45 1.18x⫾1.50 P⫽.042
2.13w⫾1.41 1.71x⫾1.31 1.84x⫾1.43 P⬍.001
619 1,326 295
2.90w⫾1.48 2.78w,x⫾1.42 2.64x⫾1.51 P⫽.032
1.30⫾1.46 1.23⫾1.47 1.25⫾1.59 P⫽.601
1.92⫾1.33 1.91⫾1.39 1.98⫾1.55 P⫽.697
a
Computer use does not include homework. SD⫽standard deviation. SES⫽socioeconomic status. w,x,y,z Means with different superscripts are significantly different (P⬍.05), means with the same superscripts are not. b c
Table 3 Sedentary behaviors (mean hours per day) by demographic characteristics in girls
Race/Ethnicity White Black Asian Hispanic Native American Mixed/Other P value SESc High Middle Low P value Age 13 years or younger 14-16 years 17 years and older P value
n
Television/Video MeanⴞSDb
Computera MeanⴞSD
Reading/Homework MeanⴞSD
1,041 406 460 110 89 100
2.35x⫾1.40 3.14y⫾1.51 2.45x⫾1.52 2.45x⫾1.35 2.66x,z⫾1.42 2.91z⫾1.45 P⬍.001
0.92z⫾1.21 1.06x,y⫾1.36 1.20x,y⫾1.43 0.82y,z⫾1.23 0.78y,z⫾1.24 1.03x,y,z⫾1.34 P⫽.001
2.41x⫾1.39 1.95y⫾1.38 2.89z⫾1.45 1.79y⫾1.30 1.87y⫾1.46 2.06y⫾1.26 P⬍.001
767 560 859
2.36x⫾1.45 2.67y⫾1.48 2.67y⫾1.47 P⬍.001
1.03⫾1.26 0.98⫾1.30 0.98⫾1.34 P⫽.688
2.61x⫾1.43 2.31y⫾1.38 2.15y⫾1.44 P⬍.001
652 1318 270
2.80x⫾1.52 2.48y⫾1.45 2.33y⫾1.46 P⬍.001
1.15x⫾1.37 0.94y⫾1.27 0.91y⫾1.26 P⫽.001
2.31⫾1.43 2.37⫾1.44 2.32⫾1.44 P⫽.715
a
Computer use does not include homework. SD⫽standard deviation. SES⫽socioeconomic status. x,y,z Means with different superscripts are significantly different (P⬍.05); means with the same superscripts are not. b c
Journal of THE AMERICAN DIETETIC ASSOCIATION / 1301
RESEARCH
Table 4 Dietary behaviors, BMI, and physical activity by each type of sedentary behavior among boys—adjusted for race, SES, age Computera
Television/Video High
Average
Low
High
Average
Reading/Homework Low
High
Average
Low
b
n⫽1279 n⫽344 n⫽533 n⫽807 n⫽900 n⫽478 n⫽1,053 n⫽709 n⫽617 4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™Mean⫾SEc™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™3 32.6⫾0.1 17.4⫾0.1 4.5⫾0.1 25.1⫾0.1 7.1⫾0.1 0.6⫾0.0 28.7⫾0.1 13.2⫾0.1 3.5⫾0.1
Total h/wk BMI (Body mass (index) 23.3x⫾0.2 Dietary behaviors Energy intake (kcal) 2,437x⫾55.0 Fat energy (% of total energy) 31.9x⫾0.0 Dietary fat (g) 41.1x⫾1.0 Soft drinks (daily servings) 1.60x⫾0.1 Fried food (daily servings) 0.66x⫾0.0 Snacks (daily servings) 3.60x⫾0.1 Fruit (daily servings) 2.14x⫾0.1 Vegetables (daily servings) 1.58x⫾0.1 Physical Activity Active Energy (kcal/kg per 66.1NS⫾1.5 week)e
23.2x,y⫾0.1
22.6y⫾0.3
2,178y⫾38.2 2,093y⫾71.2
23.1NS⫾0.2
23.0NS⫾0.2
23.2NS⫾0.2
2,350x⫾59.7
2,173y⫾48.4
2,223b⫾45.4
30.1y⫾0.0 35.4y⫾0.7
30.2y⫾0.0 32.6y⫾1.3
30.6NS⫾0.0 38.4x⫾1.1
30.5NS⫾0.0 35.6y⫾0.9
30.7NS⫾0.0 36.3x,y⫾0.8
1.39y⫾0.0
1.33y⫾0.1
1.52x⫾0.1
1.35y⫾0.0
1.47x⫾0.0
0.58y⫾0.0
0.55y⫾0.0
0.62NS⫾0.0
0.59NS⫾0.0
0.59NS⫾0.0
2.88y⫾0.1
3.06y⫾0.2
3.48x⫾0.1
2.87y⫾0.1
2.28x,y⫾0.1
2.47y⫾0.1
2.37NS⫾0.1
1.65x⫾0.1
1.81y⫾0.1
64.5NS⫾1.1
61.6NS⫾2.0
23.6x⫾0.2
23.0y⫾0.2
23.0y⫾0.2
2,231NS⫾63.3 2,227NS⫾42.8
2,248NS⫾1.0
29.9x⫾0.0 30.5x,y⫾0.0 35.5NS⫾1.2 36.5NS⫾0.8
31.2y⫾0.0 37.3NS⫾0.9
1.38x⫾0.1
1.40x⫾0.0
1.54y⫾0.0
0.57x⫾0.0 0.59x,y⫾0.0
0.63y⫾0.0
3.10y⫾0.1
2.97x⫾0.1
2.96x⫾0.1
3.41y⫾0.1
2.25NS⫾0.1
2.24NS⫾0.1
2.45x⫾0.1
2.30x⫾0.1
2.12y⫾0.1
1.76NS⫾0.1
1.63NS⫾0.1
1.62NS⫾0.1
1.92x⫾0.1
1.65y⫾0.1
1.49z⫾0.1
63.0x,y⫾1.7
67.0x⫾1.4
63.1y⫾1.3
69.5x⫾1.8
65.5x⫾1.2
59.7y⫾1.4
a
Computer use does not include homework. n reflects number of students in each level of sedentary behavior, not the final number for multivariate analyses. SE⫽standard error. d kcal⫽kilocalorie. e kcal/kg⫽kilocalories per kilogram of body weight. x,y,z Means with different superscripts are significantly different (P⬍.05); means with the same superscripts are not. NS not significant. b c
spent significantly more time than boys reading and doing homework. Mean numbers of hours per day spent with each sedentary activity by sociodemographic variables are displayed in Table 2 for boys and Table 3 for girls. Among boys, television use significantly differed by race/ethnicity and age, but not by SES. Black and Asian boys reported watching the most television, whereas Hispanic boys reported the least. Likewise, younger boys reported more time with television/videos than older boys. Among girls, television use was most common among black youth, girls in the middle- or low-SES groups, and girls age 13 and younger. Computer use was higher among Asian and white boys, and boys in the high- or middle-SES groups. There was no significant difference by age among boys reporting computer use. Asian or black girls reported the highest computer use along with girls aged 13 and younger. Unlike boys, among girls there was no significant association between computer use and SES. For both boys and girls, trends in time spent reading and doing homework were similar. Asian and white youth spent the most time in this sedentary behavior, as did youth reporting the highest SES group. Age was not significantly associated with time spent reading for either boys or girls. BMI and Sedentary Behaviors Associations between each of the sedentary behaviors and BMI, dietary behaviors, and physical activity are reported in 1302 / October 2003 Volume 103 Number 10
Table 4 for boys and Table 5 for girls. All associations have been adjusted for race/ethnicity, SES, and age. For both boys and girls, BMI was positively associated with television/video use. Boys reporting high television use had a mean BMI of 23.3, while boys reporting low television use had a mean BMI of 22.6. However, when computer use and reading/homework were added as covariates in the model, the association between television/video use and BMI was no longer significant (data not shown). Boys in the high reading group also had a significantly higher mean BMI than those reporting other levels of reading, but this association was no longer significant when the other sedentary activities were treated as covariates. Among girls, those reporting high television use and high computer use had mean BMIs (23.8) nearly one unit higher than those reporting low use of the respective activities. These findings remained significant when the other sedentary behaviors were treated as covariates. Dietary Intake and Sedentary Behaviors Dietary intake was significantly associated with most sedentary behaviors for both boys and girls. Energy intake for both boys and girls was positively associated with television/video and computer use. Boys reporting high television/video use consumed almost 400 kcal more per day than those in the low-use category. Similarly, girls reporting high computer use consumed more than 300 kcal more per day than girls reporting low use. Likewise, both boys and girls in the high television/
RESEARCH
Table 5 Dietary behaviors, BMI, and physical activity by each type of sedentary behavior among girls—adjusted for race, SES, age Computera
Television/Video Highb
Average
Low
High
Average
n⫽539
n⫽1,222
n⫽479
n⫽403
n⫽787
Reading/Homework Low
High
Average
Low
n⫽1,050
n⫽700
n⫽1,055
n⫽485
4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™Mean⫾SE ™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™3 32.2⫾0.1 16.8⫾0.1 4.4⫾0.0 23.9⫾0.1 7.0⫾0.1 0.5⫾0.0 28.8⫾0.1 14.0⫾0.1 3.8⫾0.0 c
Total h/wk BMI (Body mass index) Dietary Behaviors Energy intake (kcal)a Fat energy (% of total energy) Dietary fat (g) Soft drinks (daily servings) Fried food (daily servings) Snacks (daily servings) Fruit (daily servings) Vegetables (daily servings) Physical activity Active energy (kcal/kg per week)d
23.8x⫾0.2
23.3x,y⫾0.2
22.8y⫾0.2
24.1x⫾0.3
23.3y⫾0.2
23.1y⫾0.2
23.2NS⫾0.2
23.5NS⫾0.2
23.0NS⫾0.2
2,218x⫾59.0
1,936y⫾42.2
1,868y⫾36.2
1,991NS⫾45.3
1,924NS⫾36.6
1,963NS⫾53.9
2,118x⫾51.9
1,896y⫾33.8
1,918y⫾53.4
30.9x⫾0.0 33.3x⫾0.9
29.6y⫾0.0 28.1y⫾0.6
29.0y⫾0.0 27.9y⫾0.9
30.0NS⫾0.0 34.5x⫾1.0
29.8NS⫾0.0 29.3y⫾0.7
29.7NS⫾0.0 27.3z⫾0.6
29.0x⫾0.0 29.3NS⫾0.8
30.1y⫾0.0 29.0NS⫾0.6
30.2y⫾0.0 30.0NS⫾0.9
1.58x⫾0.1
1.27y⫾0.0
1.12z⫾0.1
1.51x⫾0.1
1.27y⫾0.0
1.27y⫾0.0
1.22x⫾0.0
1.28x⫾0.0
1.50y⫾0.1
0.64x⫾0.0
0.55y⫾0.0
0.53y⫾0.0
0.64x⫾0.0
0.54y⫾0.0
0.56y⫾0.0
0.54x⫾0.0
0.57x⫾0.0
0.62y⫾0.0
3.21x⫾0.1
2.78y⫾0.1
2.65y⫾0.1
3.43x⫾0.1
2.68y⫾0.1
2.77y⫾0.1
2.83x,y⫾0.1
2.74y⫾0.1
3.12x⫾0.1
2.24x⫾0.1
2.36x,y⫾0.1
2.57y⫾0.1
2.54x⫾0.1
2.43x,y⫾0.1
2.27y⫾0.1
2.54x⫾0.1
2.40x⫾0.1
2.09y⫾0.1
1.66x⫾0.1
1.77x⫾0.1
2.06y⫾0.1
1.98x⫾0.1
1.88x⫾0.1
1.69y⫾0.1
1.99x⫾0.1
1.78y⫾0.1
1.61y⫾0.1
45.9NS⫾1.6
48.2NS⫾1.0
49.4NS⫾1.6
52.5x⫾1.8
47.6y⫾1.3
46.4y⫾1.1
52.1x⫾1.4
47.3y⫾1.1
43.4z⫾1.6
a
Computer use does not include homework. n reflects number of students in each level of sedentary behavior, not the final number for multivariate analyses. c SE⫽standard error. d kcal⫽kilocalorie. e kcal/kg⫽kilocalories per kilogram of bodyweight. x,y,z Means with different superscripts are significantly different (P⬍.05); means with the same superscripts are not. NS not significant. b
video category were significantly more likely to consume more dietary fat and a higher percent of fat energy than the average and low users. In contrast, both boys and girls reporting high levels of time spent in reading/homework consumed significantly less percent of energy from fat than the low group. All associations remained significant when controlling for the other sedentary activities, with the exception of energy and computer use and dietary fat and computer use among boys. Both boys and girls who reported high use of television/videos consumed significantly more soft drinks, fried foods, and number of snacks per day than the average- and low-use groups. These findings contrast with findings for youth spending the most time with reading/homework. Both boys and girls who report high amounts of time spent with reading/homework consumed significantly fewer soft drinks, fried foods, and snacks per day than those in the low reading/homework category. Computer use was associated with more daily servings of soft drinks and snacks among boys and higher consumption of soft drinks, fried food, and snacks among girls. These findings all remained significant when adjusted for the other sedentary activities. High television/video use was associated with significantly
less consumption of fruits and vegetables among both boys and girls, whereas reading/homework was associated with higher consumption of fruits and vegetables. Computer use was also associated with higher consumption of fruits and vegetables for girls, but not for boys. These estimates changed only marginally and remained significant when the other sedentary behaviors were controlled for in the models. Physical Activity and Sedentary Behaviors Physical activity was not associated with television/video use for boys or girls, but was associated with computer use and time spent reading/doing homework. Boys in the average category of computer use and in the high category for reading/ homework expended more active energy than those in the low categories of the respective sedentary behavior. Among girls, those spending the most time with computers or reading/ homework reported significantly higher levels of physical activity than girls in the average and low categories for the respective sedentary activity. DISCUSSION The purpose of the current study was to describe the sociodemographic characteristics of adolescent boys and girls who Journal of THE AMERICAN DIETETIC ASSOCIATION / 1303
RESEARCH
spend time with various sedentary activities, and to describe how various levels of sedentary behaviors are associated with specific dietary behaviors, BMI, and physical activity. Boys and girls spent more time watching television/videos than with computers or reading/doing homework. On average, boys spend approximately 15 more minutes per day with television/videos than girls do. Likewise, adolescents age 13 and younger spend between 15 and 30 minutes more per day with television/videos than their older peers do. Both black boys and girls spent 3 or more hours per day with television/videos. Our findings are similar to other studies of media use among adolescents. A report by the Kaiser Family Foundation (2) also found that black youth spent more time with television than their white or Hispanic counterparts. Boys and girls spent on average twice as long with television/videos per day than with computers. Our findings are consistent with estimates of media availability in that 99% of all American teenagers live in a home with a television, whereas 73% of teenagers live in homes with a computer and less than 50% have Internet access (2).
Dietary intake was significantly associated with most sedentary behaviors for both boys and girls The most striking findings of the current research are the differences in the associations between three different sedentary activities and nutrient intake and foods consumption. High television/video use was associated with more unhealthful dietary behaviors among boys and girls, whereas those reporting inactivity due to reading/homework were more likely to report more healthful dietary behaviors. Differences in patterns of association may be due to differences in the sedentary behaviors. For example, while watching television, there are advertising influences and commercial breaks that may encourage snacking. In contrast, it would be necessary to take a break from reading/homework to snack. Alternatively, differences in patterns of association may be due to other environmental factors (eg, familial) influencing both the types of sedentary behavior children engage in as well as the types of food available at home. Differences in patterns of associations and potential explanations for these differences are discussed in more detail later in this section. Our finding that youth reporting high use of television/videos consume more energy per day is consistent with a similar crosssectional study of adolescents that found a positive association between time spent watching television and total energy intake (8). Likewise, our findings that television/video use was positively associated with BMI among boys and girls when controlling for race/ethnicity and SES are supported by several other studies (5-7). However, these previous studies did not control for other sedentary activities except videos and video viewing. In the current study, the association between BMI and television only remained significant for girls when the other sedentary activities were controlled for as covariates. This suggests that for girls, television is an independent correlate of BMI, whereas for boys there may be an association between time spent with television and other sedentary activities and BMI. 1304 / October 2003 Volume 103 Number 10
Among boys and girls, television/video use was positively associated with daily servings of soft drinks, fried foods, and snacks. Our finding may be in part explained by the role of advertising. Advertising that is targeted to children and adolescents includes mostly food and foods that are high in fat and/or sugar (28,29). Furthermore, food manufacturers allocate more than 75% of their advertising budget to television, and fast-food restaurants allocate more than 95% of their advertising budgets to television (30). Similarly, we found that computer use was positively associated with soft drinks and snack consumption among boys and girls. This may reflect the fact that food and beverage companies advertise their websites on television as well as use the Internet for drawings and sweepstakes. A study of the same population found that youth who frequented fast-food restaurants were more likely to watch television than youth who did not (19). This finding may explain the higher consumption of fried foods and soft drinks. It is of great interest that reading/doing homework was positively associated with many healthful dietary behaviors among boys and girls. To our knowledge, these associations have not been previously documented in the health and social sciences. It is also noteworthy that computer use was positively associated with fruit and vegetable consumption and physical activity among girls. Findings from a recent report showed that adolescents are likely to use the Internet for obtaining health information, particularly about chronic diseases and weight loss/ gain (31). Lastly, we did not find a significant relationship between television/video use and physical activity for either boys or girls. This may suggest that the role of television/videos in obesity may be through mediating consumption rather than energy expenditure. Other cross-sectional studies have demonstrated an inverse relationship between physical activity and television use (16-18). It is possible that we did not show similar findings due to the combination of television and video use in one variable and that our physical activity measure was self-report and based only on a three-item scale. Our study has several strengths, including a sociodemographically diverse sample of adolescents; inclusion of specific, validated measures of dietary intake; and assessment of BMI by trained research staff. Limitations include the combination of sedentary behaviors in questions and limited variety of sedentary activity measures. The study could be improved with questions asking about television and videos separately and by using more comprehensive measures of physical activity. Because our study design was cross-sectional, our findings cannot be used to determine causal pathways. Additionally, the current study only applies to youth attending school; it may be that current findings are accentuated among children who are not in school. Our findings may be further limited by other factors that may act as missing predictors (such as family and school environments) in the associations between the sedentary activities and dietary behaviors, BMI, and physical activity examined here. IMPLICATIONS The present study provides needed information for health professionals and researchers about adolescent health and nutrition. The findings suggest that messages and advice regarding reducing time spent in sedentary activities should be targeted at watching television/videos instead of time spent reading, doing homework, or using a computer. ■ Our findings also lend support to the hypothesis that the ■
RESEARCH
association between television viewing and obesity is moderated by increased consumption more than the displacement of time spent being physically active. Further research using stronger measures of physical activity may better support this hypothesis. ■ For dietitians, this study provides information about factors influencing adolescent eating behaviors and food choices. Nutrition education should incorporate messages about the influence of media and advertising on eating-related attitudes and behaviors. When providing nutrition advice about decreasing consumption of soft drinks, fast food, and snack foods, dietitians should consider the role of television in sending opposite messages. Advice about decreasing time spent being inactive should specifically target television viewing because it is so common among adolescents and is associated with poor dietary behaviors. References 1. 2000 Report on television. New York: Nielsen Media Research; 2000. 2. Roberts D, Foehr U, Rideout V, Brodie M. Kids & Media The New Millenium. Menlo Park, CA: The Henry J. Kaiser Foundation; 1999. 3. Troiano R, Kuczmarski R, Campbell S, Johnson C. Overweight prevalence trends for children and adolescents: The National Health and Nutrition Examination Surveys 1963 to 1991. Arch Pediatr Adolesc Med. 1995;149:10851091. 4. Prevalence of overweight among children and adolescents: United States, 1999. Washington, DC: National Center for Health Statistics; 2001. 5. Gortmaker S, Sobol A, Peterson K, Colditz G, Dietz W. Television viewing as a cause of increasing obesity among children in the United States, 19861990. Arch Pediatr Adolesc Med. 1996;150:356-362. 6. Dietz W, Gortmaker S. Do we fatten our children at the television set? Pediatrics. 1985;75:807-812. 7. Hernandez B, Gortmaker S, Colditz G, Peterson K, Laird N, Parra-Cabrera S. Association of obesity with physical activity, television programs and other forms of video viewing among children in Mexico City. Int J Obes Relat Metab Disord. 1999;23:845-854. 8. Crespo C, Smit E, Troiano R, Bartlett S, Macera C, Andersen R. Television watching, energy intake, and obesity in US children: Results from the Third National Health and Nutrition Examination Survey, 1988-1994. Arch Pediatr Adolesc Med. 2001;155:360-365. 9. Berkey C, Rockett H, Field A, Gillman M, Frazier A, Carmago C, Colditz G. Activity, dietary intake, and weight changes in a longitudinal study of preadolescent and adolescent boys and girls. Pediatrics. 2000;105:e56. 10. Dowda M, Ainsworth B, Addy C, Saunders R, Riner W. Environmental influences, physical activity, and weight status in 8- to 16-year olds. Arch Pediatr Adolesc Med. 2001;155:711-717. 11. Eisenmann J, Bartee R, Wang M. Physical Activity, TV Viewing, and Weight in U.S. Youth: 1999 Youth Risk Behavior Survey. Obes Res. 2002;10: 379-385. 12. Gortmaker S, Peterson K, Wiecha J, Sobol A, Dixit S, Fox M, Laird N. Reducing obesity via a school-based interdisciplinary intervention among youth: Planet Health. Arch Pediatr Adolesc Med. 1999;153:409-418. 13. Epstein L. Effects of decreasing sedentary behavior and increasing activity on weight change in obese children. Health Psychol. 1995;14:109-115.
14. Epstein L, Paluch R, Gordy C, Dorn J. Decreasing sedentary behaviors in treating pediatric obesity. Arch Pediatr Adolesc Med. 2000;154:220-225. 15. Robinson T. Reducing children’s television viewing to prevent obesity. JAMA. 1999;282:1561-1567. 16. Bungum T, Vincent M. Determinants of physical activity among female adolescents. Am J Prev Med. 1997;13:115-122. 17. Robinson T, Hammer L, Killen J, Kraemer H, Wilson D, Hayward C, Taylor C. Does television viewing increase obesity and reduce physical activity? Cross-sectional and longitudinal analyses among adolescent girls. Pediatrics. 1993;91:273-280. 18. Strauss R, Rodzilsky D, Burack G, Colin M. Psychosocial correlates of physical activity in healthy children. Arch Pediatr Adolesc Med. 2001;155:897902. 19. French S, Story M, Neumark-Sztainer D, Fulkerson J, Hannan P. Fast food restaurant use among adolescents: Associations with nutrient intake, food choices and behavioral and psychosocial variables. Int J Obes Relat Metab Disord. 2001;25:1823-1833. 20. Taras H, Sallis J, Patterson T, Nader P, Nelson J. Television’s influence on children’s diet and physical activity. J Dev Behav Pediatr. 1989;10:176-180. 21. Neumark-Sztainer D, Croll J, Story M, Hannan P, French S, Perry C. Ethnic/racial differences in weight-related concerns and behaviors among adolescent girls and boys: Findings from Project EAT. J Psychosom Res. 2002;53:963-974. 22. Rockett H, Wolf A, Colditz G. Development and reproducibility of a food frequency questionnaire to assess diets of older children and adolescents. J Am Diet Assoc. 1995;95:336-340. 23. Rockett H, Breitenbach M, Frazier A. Validation of a youth/adolescent food frequency questionnaire. Prev Med. 1997;26:808-816. 24. Diet and Health: Implications for reducing chronic disease risk. Washington, DC: National Research Council; 1989. 25. Godin G, Shepard R. A simple method to assess exercise behavior in the community. Can J Appl Sport Sci. 1985;10:141-146. 26. Sallis J, Buono M, Roby J, Micale F, Nelson J. Seven-day recall and other physical activity self-report in children and adolescents. Med Sci Sports Exerc. 1993;25:99-108. 27. Brieman L, Friedman J, Olshen R, Stone C. Classification of Regression Trees. Belmont, CA: Wadsworth International Group; 1984. 28. Kotz K, Story M. Food advertisements during children’s Saturday morning television programming: Are they consistent with dietary recommendations? J Am Diet Assoc. 1994;94:1296-1300. 29. Wilson N, Quigley R, Mansoor O. Food ads on TV: A health hazard for children? Aust N Z J Public Health. 1999;23:647-650. 30. Gallo A. Food advertising in the United States. America’s Eating Habits: Changes and Consequences. Washington, DC: USDA/ERS; 1999:173-180. 31. Rideout V. Generation Rx.com. Menlo Park, CA: The Henry J. Kaiser Foundation; 2001.
This study was supported by grant MCJ-270834 (D. Neumark-Sztainer, principal investigator) from the Maternal and Child Health Bureau (Title V, Social Security Act), Heath Resources and Service Administration, US Department of Health and Human Services. Preparation of this manuscript was supported by grant DK-50456 from the National Institute of Diabetes and Digestive and Kidney Diseases.
Journal of THE AMERICAN DIETETIC ASSOCIATION / 1305