Sociology 3 quick discussions

Disparities in Overweight and Obesity Among US College Students

Toben F. Nelson, ScD; Steven L. Gortmaker, PhD; S.V. Subramanian, PhD Lilian Cheung, ScD; Henry Wechsler, PhD

Objectives: To examine social dis- parities and behavioral correlates of overweight and obesity over time among college students. Methods: Multilevel analyses of BMI, physi- cal activity, and television viewing from 2 representative surveys of US college students (n=24,613). Results: Overweight and obesity increased over time and were higher among males, African Americans, and students of lower socioeco- nomic position and lower among Asians. Television viewing and in

activity were associated with obe- sity, and disparities in these behav- iors partially accounted for excess weight among African Americans. Conclusions: Social disparities in overweight and obesity exist among college students. Promoting physi- cal activity and reducing televi- sion viewing may counteract in- creasing trends.

Key words: obesity, college stu- dents, physical activity, televi- sion viewing, social disparities

Am J Health Behav. 2007;31(4):363-373

Overweight and obesity have in-creased dramatically over the past30 years among both adults and children in the United States.''^ The in- crease in overweight and obesity h a s been observed in all age, gender, and racial/ethnic groups^'^ and is rising more rapidly among women, young adults, His- panics and non-Hispanic blacks, and people with some college education,^'^ Higher rates are observed among minor-

Toben F. Nelson, Research Associate, Depart- ment of Society, Human Development and Health; Steven L. Gortmaker, Professor, Department of Society, Human Development and Health; S.V. Subramanian, Assistant Professor, Department of Society, Human Development and Health; Lilian Cheung, Lecturer, Department of Nutrition; Henry Wechsler, Lecturer on Society, Human Develop- ment and Health, all from the Harvard School of Public Health, Boston, MA.

Address correspondence to Dr Nelson, Harvard School of Public Health, Department of Society, Human Development and Health, 677 Hunting- ton Avenue, Boston, MA 02115. E-mail: tnelson@hsph. harvard, edu

ity racial/ethnic groups, most notably African Americans and Hispanics.*"® Per- sons of lower socioeconomic position gen- erally also have higher rates of obesity,^' Healthy People 2010 goals for the nation's health include a reduction in the preva- lence of obesity and the elimination of disparities in health across different seg- ments of the population.'"

Obesity is a s s o c i a t e d with major chronic diseases, such as cardiovascu- lar disease, some cancers, type 2 diabe- t e s , ' ' ' ^ and creates a major burden for health care systems.'^'''^ Although the full population health consequences of this epidemic have not yet been realized, the potential impact for future decreased life expectancy and poor health due to obesity is considerable.'* The poor health out- comes of obesity usually manifest in the later stages of life, but their causes can develop in childhood or young adulthood. The transition from adolescence to adult- hood is one developmental period that may be a critical stage for weight gain. Body mass index (BMI) in early adulthood is an important predictor for subsequent

Am J Health Behav.™ 2007;31(4):363-373 363

Disparities in Overweight and Obesity

obesity." It is also during the young adult- hood period when social patterning in obesity emerges strongly.^"

In A Call to Action to Prevent and De- crease Overweight and Obesity, the US Surgeon General recommends schools as an important setting in which to address overweight and obesity.'^ To date, most school-based research and intervention activity has focused on primary and sec- ondary schools. The college setting pre- sents an important opportunity for health promotion during a critical stage of devel- opment. One in 3 young adults attend college.̂ ^ However, few studies have ex- amined the prevalence and patterns of overweight and obesity among college stu- dents, and no studies have examined whether social disparities in overweight and obesity exist among college students.

Greater caloric intake than expendi- ture leads to overweight,'^ and specific behavior targets to prevent and reduce excess weight include diet and physical activity.'̂ '̂ '̂̂ ^ Television viewing also ap- pears to have effects on overweight inde- pendent from inactivity.^^ Television view- ing is associated with exposure to food and beverage advertising and with be- tween-meal snacking.^*"^^ Intervention studies have shovwi that reducing televi- sion viewing leads to reductions in over- weight and obesity in children.^^"^' The prevalence of these behaviors and their relationship to overweight and obesity h a s not been systematically studied among college students.

This study is the first to examine preva- lence, trends and social disparities in overweight, obesity, and class II obesity in a nationally representative sample of college students in the United States. Although other studies have examined overweight and obesity in this population, only one obtained a nationally represen- tative sample of students.^" The present study has sufficient sample size to exam- ine differences among different groups of s t u d e n t s , and it was administered in multiple years to track changes over time. We test the hypothesis that the preva- lence of overweight and obesity increased among college students in the United States from 1993 to 1999. We also exam- ine disparities, or inequalities in over- weight and obesity defined by sex, race/ ethnicity, socioeconomic position and age in this representative sample of college s t u d e n t s . We hypothesize that higher

rates of overweight and obesity occur among males compared with females, m e m b e r s of minority r a c i a l / e t h n i c groups, students of lower socioeconomic position, and upper class (by year in school) compared with underclass students. The association of overweight, obesity, and class II obesity with television viewing and physical activity, whether these rela- tionships are consistent across student subgroups, whether they change over time, and if they account for increasing body weight or social disparities in these measures are examined.

METHODS Sample Data were from the Harvard School of

Public Health (HSPH) College Alcohol Study (CAS), a nationally representative sample of students attending 4-year colleges in the United States. Colleges were selected proportionate to the size of the school from a list of all 4-year institutions pro- vided by the American Council on Educa- tion. The sample for the present analysis included 119 colleges that had data in both the 1993 and 1999 surveys, consis- tent with previous analyses of these data.^' Students were 225 full-time undergradu- ates randomly sampled within each col- lege. The registrar at each participating school was provided instructions on draw- ing a random sample of full-time stu- dents. For the present analysis the sample was limited to 24,613 students (12,786 in 1993 and 11,827 in 1999) under 25 years of age (mean = 20.4; s.d. = 1.6). Although these surveys were administered 6 years apart there is the potential that the same students could have responded to both, thereby reducing the variation in the sample. We did not assess this potential between the 1993 and 1999 surveys. However, in other administrations of the CAS we found an overlap of 1.50% for a 2- year difference between surveys and 0.46 % for a 4-year difference between sur- veys. We have found no statistical evi- dence of reduced variation resulting from the inclusion of these respondents.^^

The 1993 and 1999 administrations of the CAS collected data on exact height and weight. Response rate was 70% in 1993 (range at each college was 41 to 100%) and was 60% in 1999 (range 40- 83%). The correlation between response rate and body mass index at the college level was r =-0.18 (N=119; P=0.05) in 1993

364

Nelson et al

and r = -0.16 (N=119; P=0.08) in 1999. All analyses were adjusted for college re- sponse rate to account for response bias, although it was not statistically signifi- cant and did not alter the results. Models stratified by high and low response rates showed similar r e s u l t s . Data were weighted to match each school's true demographic characteristics over 8 strata of gender, 2 age-groups (<22 vs others) and 2 ethnic groups (white vs others). Additional details of the study methodol- ogy and sampling procedure are published elsewhere.^'

Measures Respondents reported current height

in feet and inches and weight in pounds. Self-report measures of height and weight are generally considered to be valid and reliable for large-scale surveillance sur- veys.̂ ^ Three measures were calculated based on body mass index (BMI), expressed in kilograms of body weight per meters of height squared (kg/m^): (a) overweight (BMI >= 25 kg/m^), (b) obesity (BMI >= 30 kg/m^), and (c) class II obesity (BMI >= 35 k / = ) 3 *

Respondents described the racial/eth- nic group they belonged to using the fol- lowing categories: white; black/African American; Asian/Pacific Islander; Na- tive American Indian/Native Alaskan; Other. The Native American and other race categories were combined due to small cell sizes for each. Hispanic origin was included in a separate question and was modeled separately. Socioeconomic position (SEP) was assessed as educa- tional attainment for each parent and converted into a 3-level variable in which (a) neither parent attended college; (b) one, but not both, parents attended col- lege; and (c) both parents attended col- lege, consistent with previous analysis of these data.^^ Students with missing data on these variables of interest occurred in less than 1% of the cases, and these were excluded from the analysis.

Respondents were asked a series of questions about the amount of time per day on average they spend on each of 9 different activities, including one ques- tion about television viewing and 2 ques- tions about physical activity. Television viewing was measured as the average number of hours per day, ranging from zero to 5 or more. Physical activity was defined as any participation in intercolle-

giate athletics or other physical activity (yes vs no). An additional 348 subjects (.01% of the analytic sample) were miss- ing data for activity, and 97 (.004%) were m i s s i n g d a t a for television viewing. These subjects were deleted from analy- ses examining television viewing and activity, and nested models were com- pared only for those respondents with complete data.

Analysis Descriptive analyses and cross-tabula-

tions were conducted in SAS version 9.0 on the UNIX platform (The SAS Institute, Inc., Cary, NC). Multilevel analytic tech- niques were used in a 2-level framework (college and individual) to account for the clustered sampling scheme in MLwiN software version 2.0.^* Change over time in each of the 3 outcome variables was assessed using an indicator variable for survey year adjusting for student gender, race/ethnicity, SEP, and year in school. Change over time in each group was examined using interaction terms be- tween survey year and gender, r a c e / ethnicity, and SEP. Differences in out- come variables for gender over time and by race/ethnicity were observed, so sub- sequent analyses were stratified by gen- der. In gender-stratified models, interac- tion terms for race/ethnicity with SEP examined whether SEP modified the as- sociation between r a c e / e t h n i c i t y and overweight or obesity. Whether Hispanic subgroups differed by race was also exam- ined.

Gender-stratified analyses were used to test for differences between student subgroups by television viewing and physi- cal activity. Television viewing and physi- cal activity variables were added to each analysis to examine the relationship of these variables with overweight and obe- sity and to determine whether the addi- tion of these variables attenuated the differences in prevalence of the other variables. To examine whether the rela- tionship of television viewing and physi- cal activity differed by population groups, similar models were stratified by race/ ethnicity.

Multilevel logistic regression models were fitted using the logit-link function for binomial outcomes, second-order pe- nalized quasi-likelihood and iterative generalized least s q u a r e s procedures. MLwiN employs a Taylor series lineariza-

Am J Health Behav.™ 2007;31(4):363-373 365

Disparities in Overweight and Obesity

Table 1 Prevalence of Overweight, Obesity and Class II

Socio-demographic Characteristics

Gender Female

Male

Race/ethnicity White

African American Asian

Native American/Other Hispanic

Socioeconomic Position Both Parents Attended College

One Parent (not both) Attended College Neither Parent Attended College

Year in School First year

Sophomore Junior Senior

Fifth year

Sample Size 1993

7369 5417

10,624 568 849 745 733

7454 3256 2076

2864 2648 3110 3045 1119

1999

7258 4569

9307 633 978 909 743

7412 2841 1574

2993 2845 2912 2382

695

Overweight (BMI 1993

13.5 30.8

21.5 33.3 13.6 23.9 25.0

20.4 23.4 23.5

18.9 19.5 22.0 23.8 28.7

>=25) 1999

20.1 35.0

26.7 38.3 16.4 30.6 30.2

25.0 29.1 31.4

23.0 27.3 27.6 27.5 37.2

Obesity by

Obesity Class II obesity (BMI 1993

2.9 5.4

3.9 11.2 2.0 3.4 2.8

3.6 4.6 4.9

3.1 4.0 4.3 4.3 5.7

>=30) 1999

5.4 7.8

6.2 13.9 2.3 8.2 8.3

5.9 7.6 7.2

5.2 6.7 7.2 5.7

10.9

(BMI>=35) 1993

1.0 0.8

0.7 4.4 0.2 0.6 0.4

0.7 1.2 1.2

0.7 0.8 0.9 1.0 1.2

1999

2.0 1.8

1.7 5.3 0.6 2.1 2.2

1.6 2.0 2.9

1.7 2.5 1.5 1.3 3.7

tion of the discrete response outcome and appropriately estimates standard errors within the multilevel clustered sampling design.^* Models were specified to account for college-level variation for each survey year. The gender-stratified models for the class II obesity outcome did not converge under these specifications so a first-or- der procedure was employed. The analy- ses using television viewing as the out- come used a normal distribution and the identity link function. The analyses strati- fied by race/ethnicity were conducted in SAS using the generalized estimating equation (GEE) estimating approach and the GENMOD procedure.

RESULTS Overweight rose significantly from

21.7% in 1993 to 26.8% in 1999, adjusting for gender, race/ethnicity, SEP, and year in school (adjusted odds ratio 1.33, 95% confidence interval 1.21-1.46, P<0.001). Similar increases were noted for obesity (4.1% in 1993 to 6.5% in 1999; AOR 1.64,

95% CI 1.39-1.93, P<0.001) and class II obesity (0.9% in 1993 to 1.9%; AOR 1.71, 95% CI 1.27-2.30, P<0.001). Changes were noted only in weight, whereas height remained stable.

Overweight, obesity, and class II obe- sity increased significantly from 1993 to 1999 in all groups, but rates differed by gender, race/ethnicity, SEP, and year in school (Table 1). Males were significantly more likely to be overweight and obese. However, there was no significant differ- ence by student gender for class II obe- sity. In gender-stratified analyses, sig- nificant differences emerged by r a c e / ethnicity, SEP, and year in school (Table 2). Among male racial/ethnic groups, over- weight was more prevalent among Afri- can Americans and Hispanics and less common among Asians compared with whites. Among females, similar racial/ ethnic differences emerged, although no differences existed between Hispanic and white females. Students of lower SEP had higher rates of overweight. Higher preva-

366

Nelson et al

Table 2 Relationship Between Overweight, Obesity, and Class II Obesity

and Socio-demographic Characteristics, Stratified by Gender

Overweight (BMI >=25) Male Female

Obesity (BMI>=30) Male Female

Class II obesity (BMI>=35) Male Female

Year 1993 1999

Year in School First year

Sophomore Junior Senior

5th year

1.00 1.16(1.03, 1.31)'

1.00 1.18 (rO2, 1.36)'

1.36(1.18, 1.56)"' 1.44(1.25, 1.66)*" 1.95(1.60, 2 . 3 7 ) " '

Race/ethnicily White 1.00

African American 1.48(1.15,1.90)'" Asian 0.59 (0.48, 0 . 7 3 ) ' "

Native American/Other 0.91 (0.71, 1.17) Hispanic 1.27(1.04,1.56)'

Socioeconomic Position Both Parents

Attended College 1.00 One Parent (not both)

Attended College 1.18(1.05, 1.32)"

Neither Parent

Attended College 1.17(1.03,1.34)'

1.00 1.51(1.31, 1 . 7 5 ) ' "

1.00 1.09(0.95, 1.25) 1.05(0.91, 1.21) 1.16(1.01, 1.34)' 1.42(1.14, 1.77)"

1.00 2 . 3 2 ( 1 . 9 4 , 2 . 7 8 ) ' " 0 4 7 ( 0 . 3 7 , 0 . 6 1 ) ' " 1.38(1.10, 1.75)" 0.98(0.76, 1.26)

1.00 1.42(1.11, 1.82)"

1.00 1.16(0.88, 1.53)

1.40(1.06, 1.84)" 1.26(0.95, 1.66)-

1.97(1.38,2.79)'"

1.00 2 . 2 2 ( 1 . 4 6 , 3 . 3 5 ) ' " 0.54 (0.30, 0 . 9 7 ) "

1.11(0.77, 1.36) 1.02(0.77, 1.56)

1.00 1.00

1.22(1.08, 1.37)" 1.16(0.88, 1.53)

1.45(1.26, 1 . 6 7 ) ' " 1.02(0.77, 1.36)

1.00 1.99(1.51,2.60)"

1.00 1.44(1.10, 1.90)" 1.41 (1.05, 1.89)' 1.33(0.96, 1.84)~

2 . 0 0 ( 1 . 3 1 , 3 . 0 6 ) " '

1.00 3 . 2 2 ( 2 . 3 8 , 4 . 3 5 ) " ' 0 . 3 5 ( 0 . 1 8 , 0 . 6 5 ) ' "

1.44(0.89,2.35) 0.85(0.50, 1.44)

1.00

1.32(1.08, 1.62)"

1.67(1.24,2.25)'"

1.00 1.88(1.16,3.04)'

1.00 1.05(0.60, 1.83) 1.07(0.62, 1.84) 1.47(0.86,2.53)

2 . 5 9 ( 1 . 4 1 , 4 . 7 7 ) "

1.00 3 . 5 7 ( 2 . 1 3 , 5 . 9 9 ) ' "

0.76(0.35, 1.67) 1.37(0.63,2.99) 0.80(0.33, 1.97)

1.00

1.40(0.93,2.11)

1.30(0.78,2.15)

1.00 1.49(0.99,2.25)-

1.00 1.63(1.08,2.45)' 1.25(0.77,2.04) 0.96(0.60, 1.53) 2.01 (1.05, 3.84)'

1.00 3 . 5 7 ( 2 . 3 8 , 5 . 3 6 ) " ' 0.08(0.01,0.65)"

1.36(0.55,3.35) 0.76(0.32, 1.82)

1.00

1.16(0.80, 1.67)

1.90(1.29,2.80)"'

Note. ~P<.10; *P<.05; **P<.01; ***P<.001

lence rates of overweight were observed in successive years in school. Neither racial/ethnic differences by SEP nor ra- cial differences among subgroups of His- panics were observed. A similar pattern in the trends and differences between student subgroups for obesity and class II obesity emerged.

Student Behaviors Approximately 3 in 4 students reported

engaging in some form of moderate or vigorous physical activity in both 1993 (75%) and in 1999 (74%). Males were more likely to be physicaJly active com- pared with females (80% vs 70% in 1993; P<0.0001; and 78% vs 7 1 % in 1999; P<0.0001). Students reported watching an average of 2 hours of television per day, and this did not differ for males and females. There was no significant differ- ence in reported physical activity or tele- vision viewing by survey year. African American students reported more televi- sion viewing than did other racial/ethnic groups among males (2.8 hours per day; s.d. = 1.7 compared with 2.1 hours overall; s.d. = 1.4; AOR 2.85; 95% CI 1.24 - 6.54).

Students of lower SEP watched signifi- cantly more television among males (2.3 hours; s.d. = 1.5; AOR 1.91; 95% CI 1.19 - 3.07 for neither low SEP, and 2.2 hours; s.d. = 1.5; AOR 1.68; 95% CI 1.05-2.71 for mid SEP compared with high SEP). No significant differences in physical activ- ity by race/ethnicity or SEP were ob- served for physical activity among males or females. Although the average num- ber of hours of television viewing among African American females was also higher (2.9 hours per day; s.d. = 1.6 compared with 1.9 hours overall; s.d. = 1.3), these results were not statistically significant (AOR 2.03; 95% CI 0.87 - 4.72). Among females, physical activity was less preva- lent for African Americans (55.6% com- pared with 70.9 % overall; AOR 0.50; 95% CI 0.43 - 0.59) and Asians (62.9%; AOR 0.63; 95% CI 0.54 - 0.74) and students of low SEP (64.2%; AOR 0.76; 95% CI 0.67 - 0.86 for low SEP; 68.4%; AOR 0.85; 95% CI 0.79 - 0.93 for mid SEP compared with high SEP, 73.1%).

The association of overweight and obe- sity with both physical activity and televi- sion viewing was examined in gender-

Am J Health Behav.™ 2007;31(4):363-373 367

Disparities in Overweight and Obesity

Prevalence of

Male

Female

Tahle 3 Overweight, Ohesity and Class II Ohesity hy

Survey Year, Gender, Physical Activity, and Average

Overall

1993 1999 1993 1999

% Overweight Male

Female

% Obese Male

Female

% Class Male

Female

1993 1999 1993 1999

1993 1999 1993 1999

II Obesity 1993 1999 1993 1999

30.8 35.0 13.5 20.1

5.4 7.8 2.9 5.4

0.8 1.8 1.0 2.0

Daily Television Viewing

by Physical Activity

Active

79.7 78.1 70.4 70.4

31.0 34.4 12.5 18.2

4.9 4.2 2.4 6.9

0.7 1.5 0.6 1.2

Not Active 0

20.3 21.9 29.6 29.6

30.0 37.5 15.8 24.9

7.2 8.3 4.1

10.9

1.0 2.8 1.8 3.9

by

hrs.

11.3 11.4 13.7 14.2

22.3 26.1

9.4 13.5

2.2 4.4 1.4 1.9

0.6 1.0 0.5 1.0

Average Daily

l h r . 2

29.0 29.2 28.9 32.2

26.8 32.9 11.3 17.5

3.2 5.8 2.1 3.9

0.3 0.9 0.8 1.3

Television

-hrs 3

26.0 26.8 25.7 24.9

30.2 35.7 13.3 20.6

5.1 7.1 2.8 6.8

0.6 2.0 0.9 1.1

Viewing

hrs. 4+

16.4 17.0 15.9 14.6

35.0 39.9 14.9 23.6

6.9 9.5 3.3 6.0

0.8 2.0 0.7 1.9

hrs.

17.4 15.6 15.8 14.1

40.0 39.5 19.2 28.3

10.1 13.6 5.2 9.2

2.0 3.4 2.1 3.4

stratified analyses adjusting for survey yeeir, race/ethnicity, and SEP. Televi- sion viewing was positively associated with overweight among males (AOR 1.14, 95% CI 1.11 - 1.18) and females (AOR 1.16, 95% CI 1.12 - 1.20). Similar findings were observed for obesity (AOR 1.31, 95% CI 1.24 - 1.39 for males and AOR 1.22, 95% CI 1.14 - 1.29 for females) and class II obesity (AOR 1.31, 95% CI 1.16 - 1.48 for males and AOR 1.16, 95% CI 1.03 - 1.32 for females). Students who engaged in physi- cal activity were less likely to be over- weight among females (AOR 0.75, 95% CI 0.68 - 0.84, adjusting for television view- ing), but not among males. Physically active students were less likely to be obese among both females (AOR 0.56, 95% CI 0.47 - 0.67) and males (AOR 0.63, 95% CI 0.52 - 0.76) and were less likely to meet criteria for class II obesity among both females (AOR 0.33, 95% CI 0.23 - 0.46) and males (AOR 0.59, 95% CI 0.40 -

0.87). The relationships of physical activ- ity and television viewing with overweight, obesity, and class II obesity were similar in each survey year and were consistent with overall findings across racial/eth- nic groups. No significant interaction effects for any of the race/ethnicity cat- egories with physical activity or televi- sion viewing for overweight, obesity, and class II obesity in the gender stratified multilevel models were observed.

Adding the physical activity and televi- sion viewing variables to the analyses improved the overall model fit and re- duced the odds of overweight for African Americans in the model for females (from AOR = 2.32; 95% CI 1.94 - 2.78 to AOR = 1.93; 95% CI 1.61 - 2.33) and for males (from AOR = 2.32; 95% CI 1.94 - 2.78 to AOR = 1.93; 95% CI 1.61 - 2.33), although African Americans remained at signifi- cantly higher odds of being overweight in nested models of students with complete

368

Nelson et al

data. Physical activity and television view- ing did not appear to account for other d i s p a r i t i e s in overweight by r a c e / ethnicity or SEP, or changes in the preva- lence of overweight over time. Similar results for obesity were observed, with declines in the odds for African American females from AOR = 3.25 (95% CI 2.55 - 4.15) to AOR = 2.49 (95% CI 1.93 - 3.20) and for African American males from AOR = 1.98 (95% CI 1.42 - 2.75) to AOR = 1.69 (95% CI 1.21 - 2.37) when accounting for physical activity and television viewing, but no other shifts in odds for any of the other groups. Similar declines in the odds of class II obesity for both female (from AOR = 3.56; 95% CI 2.37 - 5.34 to AOR = 2.59; 95% CI 1.68 - 3.98) and male (from AOR = 3.48; 95% CI 2.00 - 6.04 to AOR = 2.96; 95% CI 1.69 - 5.19) African American s t u d e n t s were observed. No other changes were noted when account- ing for physical activity and television viewing in the models.

DISCUSSION A substantial number of college stu-

dents are overweight and the prevalence, like that in the US population overall,''^'^' has increased over time. Class II obesity rose rapidly during the study period, which is particularly concerning considering the substantial health care costs associated with extreme obesity.'̂ ••'̂

Male college students were more likely to be overweight and obese than their female peers, consistent with findings in other surveys of youth and adults.'•^•^°'^'''° However, there were no differences in class II obesity between males and fe- males. Females report greater concern about their weight and may expend more effort to maintain or lose weight through dietary control and exercise.^" Socioeco- nomic position was associated with over- weight, obesity, and class II obesity among both males and females, although the strength of this association was stronger among females. These results are con- sistent with findings across multiple age- groups that show differences in rates of overweight and obesity by gender and socioeconomic position."" People of higher socioeconomic position tend to have greater awareness of weight and expend more effort at maintaining weight through dietary control and physical activity, which may explain these differences.'""''^ Fe- males could also be more likely to be

accepted into college if they are thin or of normal weight, compared with males. A third factor that promotes both academic achievement and interest in weight con- trol may account for lower rates of over- weight among females. These social forces apply to both males and females, although they may apply differentially to males. Females may also experience disproportionate weight-related discrimi- nation compared with males and experi- ence negative social consequences as a result.""

Higher rates of overweight and obesity were observed among male and female African American s t u d e n t s , consistent with other studies.2''° These racial/eth- nic disparities emerged in a sample ad- justed for parent socioeconomic position and restricted to college students, itself an indicator of higher socioeconomic po- sition. We found lower rates of overweight and obesity among Asian students. How- ever, the relationship between BMI and cardiovascular disease may vary by ra- cial/ethnic group."5"^ Prevalence rates for overweight in both survey years and obesity in the 1999 survey year were higher among Hispanics compared with whites, but these differences were not significant when adjusting for socioeco- nomic position. These findings are differ- ent from those observed in other stud- ies'''^'' and may refiect socioeconomic dif- ferences. In the CAS sample, more His- panic students reported that neither par- ent attended college (27%) compared with students overall (14%).

Significantly higher rates of overweight and obesity occurred among students in their later years of college in both sur- veys. However, these data are cross- sectional, and h y p o t h e s e s about the course of weight gain during college can- not be adequately addressed with this design. Longitudinal studies tracking in- dividuEil students throughout their col- lege years may provide more insight into the factors that promote healthy weight m a i n t e n a n c e . One l i m i t a t i o n of the present study was that the CAS was ad- ministered in the spring of the school year and may have missed significant weight gain occurring upon entry to col- lege.

Late adolescence and early adulthood may be a time of particular risk for weight gain beyond normal and healthy develop- ment. Data from the CDC Youth Risk

Am J Health Behav.™ 2007;31(4):363-373 369

Disparities in Overweight and Obesity

Behaviors Survey show that in 1999, ap- proximately 12% of male and 8% of female high school students in the United States were overweight.''* Among young adults aged 1 8 - 2 4 years, 37% of males and 21% of females were overweight in 1993, and these estimates rose to 42% for males and 28% for females in 1999.''̂ '5° The lower overall prevalence rates in the present study compared with the BRFSS and with data from the National College Health Risk Behavior Survey may reflect the higher socioeconomic position of young adults who attend college compared to those who do not attend college and vnth students who attend 4-year colleges, com- pared with students who attend all types

g Physical activity, diet, and television

viewing are 3 important behavioral tar- gets to prevent or counteract overweight and obesity.'^ Females who were active were less likely to be overweight, whereas there was no significant relationship be- tween physical activity and overweight among males. Both males and females who engaged in moderate or vigorous activity were less likely to be obese. Hours of television vievwng were strongly asso- ciated with being overweight, consistent with findings in other populations,^''^^ and may reflect the role of heavy advertising of calorie-dense, low-nutrient foods on television.^'' The average college student reports watching 2 hours of television per day. African American students reported watching an average of nearly 3 hours of television per day. Considering findings from other studies that show reducing television viewing prevents weight gain, '̂'" ^' this is a promising behavioral target for preventing overweight and obesity among college students.

Activity or television-viewing behav- iors did not change over time and are not likely to be driving the changes in body weight in the college student population over time. They also do not appear to account for between-group differences in body mass. Adjusting for these behaviors reduced the magnitude of the observed disparities in overweight among African American males and females, suggesting that those behaviors may account for some racial disparities in overweight and obesity. However, adding the activity and television-viewing v a r i a b l e s did not eliminat disparities in this group and did not change estimates for the other ra-

cial/ethnic or socioeconomic groups. It is possible that the content of the televi- sion viewed differed for some groups com- pared with others and these exposure differences could account for the remain- ing social disparities in overweight and obesity. For example, television program- ming viewed more frequently by African American students may contain more advertisements for calorie-dense fast food, promoting its consumption. Consuming fast food is associated with increased intake of overall calories, dietary fat, car- bohydrate, added sugars, and sugar-sweet- ened beverages among children.^^ Future research among college students could benefit from increased measurement pre- cision and longitudinal research design.

Some caveats and potential limitations should be considered when examining this evidence. Self-report height and weight are subject to reporting bias Eind may lead to lower estimated BMI com- pared to measured height and weight.̂ '̂̂ "* As a result, our findings may reflect lower rates of overweight and obesity than the true prevalence of these conditions in the population. In addition, females are more likely to underreport their weight com- pared with males, and this may at least partly explain the differences we observed between males and females.̂ '̂̂ * However, self-reported heights and weights are gen- erally considered valid and can be reliably used in large-scale national survey re- search, particularly with young adults.^^ In addition, we did not have measures of height and weight prior to students enter- ing college to examine changes during college. The measures of physical activ- ity and television viewing were limited to single questions and may not provide a complete reflection of students' behavior. The imprecision of these variables may have reduced our ability to estimate the true prevalence in the college student population and to examine the complex relationship vwth overweight and obesity by student subgroups. However, physical activity and television viewing were as- sociated with overweight and obesity in this sample, consistent with existing lit- erature,^^'^'' and the prevalence rates for these measures were consistent across survey years. Nonresponse bias may have influenced the results. However, body mass index was not correlated with sur- vey response rate, the results did not differ when adjusting for college response

370

Nelson et al

rate, and models stratified by high and low response rate were similar. Prevalence rates were consistent with other stud- jgg 39,40 Possible mechanisms not tested in this study include dietary behavior, self-perceived weight, ideal body size, dis- crimination, and social isolation. We did not collect dietary consumption data, and this represents an area for future re- search £ind prevention efforts in the col- lege student population.

Young people in the transition from adolescence to adulthood are at risk for excess weight gain. Students entering college may be making independent deci- sions about their diet, activity, and televi- sion viewing behaviors for the first time. New environmental and social factors may emerge during this time period to have a greater influence on their behav- ior.̂ ^ Understanding these influences on student behavior and the environments in which they live may help prevent the epidemic of overweight and obesity. Addi- tional studies should investigate poten- tial mechanisms that may create social disparities in overweight and obesity among college s t u d e n t s . Future work should also consider to what extent these mechanisms are unique to the college setting or whether some aspects of col- lege life may be amenable to change to reduce overweight and obesity. The present study suggests that reducing tele- vision viewing and increasing physical activity may help counteract overweight and obesity, but they are not likely the fundamental causes of the increase in prevalence we observed in this study. Additional s t u d y u s i n g m e a s u r e s of height, weight, television viewing, and physical activity that allows greater pre- cision may help inform these questions. Dietary intake and the food environment at college are also potential topics for further investigation. Parents, college administrators and staff, peers, and other health professionals can engage students in u n d e r s t a n d i n g the components of m a i n t a i n i n g h e a l t h y weight and can shape environments so students are more likely to engage in healthy behaviors that affect their weight.

Acknowle dgment This research was supported by grants

from the Robert Wood Johnson Founda- tion to H. Wechsler and by the Centers for Disease Control and Prevention, Preven-

tion Research Centers. •

REFERENCES l.Mokdad AH, Bowman BA, Ford ES, et al. The

continuing epidemics of obesity and diabetes in the United States. JAMA. 2001;286(10):1195-1200.

2.Hedley AA, Ogden CL, Johnson CL, et al. Prevalence of overweight and obesity among US children, adolescents, and adults, 1999- 2002. JAMA. 2004;291(23):2847-2850.

3.Mokdad AH, Serdula MK, Dietz WH, et al. The spread of the obesity epidemic in the United States, 1991-1998. JAMA. 1999;282(16):1519- 1522.

4.Ogden CL, Fiyar CD, Carroll MD, Flegal KM. Mean body weight, height, and body mass index. United States 1960-2002. Advance data from vital and health statistics; no. 347. Hyattsville, Maryland: National Center for Health Statistics;2004.

5.Truong KD, Sturm R. Weight gain trends across sociodemographic groups in the United States. Am J Public Health. 2005;95(9):1602- 1606.

6.Flegal KM, Carroll MD, Kuczmarski RJ, et al. Overweight and obesity in the United States: prevalence and trends, 1960-1994. Int J Obes Relat Metab Disord. 1998;22(l):39-47.

7.Ogden CL, Flegal KM, Carroll MD, et al. Prevalence and trends in overweight among US children and adolescents, 1999-2000. JAMA. 2002;288(14): 1728-1732.

8.Troiano RP, Flegal KM. Overweight children and adolescents: description, epidemiology, and demographics. Pediatrics. 1998; 101(3 Pt 2):497-504.

9.Moore DB, Howell PB, Treiber FA. Changes in overweight in youth over a period of 7 years: impact of ethnicity, gender and socioeco- nomic status. Bthn Dis. 2002;12(l):Sl-83-86.

lO.U.S.D.H.H.S. Healthy People 2010: Leading Health Indicators (on-line). Available at: http:/ /www.healthypeople.gov/. Accessed Decem- ber 2, 2004.

11.Flegal KM, Graubard BI, Williamson DF, et al. Excess deaths associated with underweight, overweight, and obesity. JAMA. 2005;293(15):1861-1867.

12.Calle EE, Rodriguez C, Walker-Thurmond K, et al. Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. adults. N Engi J Med. 2003;348(17):1625- 1638.

13.Field AE, Coakley EH, Must A, et al. Impact of overweight on the risk of developing common chronic diseases during a 10-year period. Arch Intern Med. 2001;161(13):1581-1586.

14.Hu FB, Manson JE, Stampfer MJ, et al. Diet, lifestyle, and the risk of type 2 diabetes mellitus in women. N Engi J Med. 2001;345(ll):790-797.

15.U.S.D.H.H.S. The Surgeon General's Call to Action to prevent and decrease overweight and obesity (on-line). Available at: http://

Am J Health Behav.™ 2007;31(4):363-373 371

Disparities in Overweight and Obesity

www.surgeongeneral.gov/topics/obesity/ calltoaction/CalltoAction.pdf. Accessed Janu- ary 11, 2005.

16.Finkelstem EA, Fiebelkorn IC, Wang G. State- level estimates of annual medical expendi- tures attributable to obesity. Obes Res. 2004,12(1): 18-24.

17.Wolf AM, Colditz GA. Social and economic effects of body weight in the United States. Am J Clin Nutr. 1996;63(Suppl 3):466S-469S.

18.01shansky SJ, Passaro DJ, Hershow RC, et al. A potentied decline in life expectancy in the United States in the 21st century. N Engl J Med. 2005;352(ll): 1138-1145.

19.Gordon-Larsen P, Adair LS, Nelson MC, et al. Five-year obesity incidence in the transition period between adolescence and adulthood: the National Longitudinal Study of Adolescent Health. Am J Clin Nutr. 2004;80(3):569-575.

2O.McTigue KM, Garrett JM, Popkin BM. The natural history of the development of obesity in a cohort of young U.S. adults between 1981 and 1998. Ann Intern Med. 2002;136(12):857- 864.

21.Center for Urban Education, Rossier School of Education, University of Southern California (on-line). Available at: http://www.usc.edu/ dept/education/CUE/statsatglance.htm. Ac- cessed February 22, 2005.

22.Blair SN, Brodney S. Effects of physical inactivity and obesity on morbidity and mor- tality: current evidence and research issues. Med Sai Sports Exerc. 1999;31(Suppl 11):S646- S662.

23.Ching PL, Willett WC, Rimm EB, et al. Activ- ity level and risk of overweight in male health professionals. Am J Public Health. 1996;86(l):25-30.

24.Wiecha JL, Peterson KE, Ludwig DS, et al. When children eat what they watch: impact of television viewing on dietary intake in youth. Arch Pediatr Adolesc Med. 2006;160(4):436- 442.

25.Boynton-Jarrett R, Thomas TN, Peterson KE, et al. Impact of television viewing patterns on fruit and vegetable consumption among ado- lescents. Pediatrics. 2003;112(6 Pt 1):1321- 1326.

26.Ludwig DS, Gortmaker SL. Programming obe- sity in chUdhood. Lancet. 2004;364(9430):226- 227.

27.Robinson TN. Reducing children's television viewing to prevent obesity: a randomized controlled trial. JAMA. 1999;282(16):1561- 1567.

28.Gortmaker SL, Peterson K, Wiecha J, et al. Reducing obesity via a school-based interdis- ciplinary intervention among youth: Planet Health. Arch Pediatr Adolesc Med. 1999;153(4):409-418.

29.Epstein LH, Paluch RA, Gordy CC, et al. Decreasing sedentary behaviors in treating pediatric obesity. Arch Pediatr Adolesc Med. 2000;154(3):220-226.

30.Lowry R, Galuska DA, Fulton JE, et al.

Physical activity, food choice, and weight management goals and practices among US college students. AmJPrevMed. 2000;18(l):18- 27.

31.Wechsler H, Lee JE, Kuo M, et al. CoUege binge drinking in the 1990s: a continuing problem. Results of the Harvard School of Public Health 1999 College Alcohol Study. J Am Coll Health. 2000;48(5): 199-210.

32.Wechsler H, Lee JE, Kuo M, et al. Trends in college binge drinking during a period of increased prevention efforts. Findings from 4 Harvard School of Public Health College Alco- hol Study surveys: 1993-2001. JAm Coll Health. 2002;50(5):203-217.

33.Kuczmarski MF, Kuczmarski RJ, Najjar M. Effects of age on validity of self-reported height, weight, and body mass index: find- ings from the Third National Health and Nutrition Examination Survey, 1988-1994. J Am Diet Assoc. 2001;101(l):28-34.

34.National Heart, Lung, and Blood Institute, National Institutes of Health. Clinical Guide- lines on the identification, evaluation, and treatment of overweight and obesity in adults. Public Health Service 1998.

35.Weitzman ER. Poor mental health, depres- sion, and associations with alcohol consump- tion, harm, and abuse in a national sample of young adults in college. J Nerv Ment Dis. 2004;192(4):269-277.

36.Rasbash J, Steele F, Browne W, et al. A User's Guide to MLwiN, Verson 2.0. Centre for Multilevel Modelling, Institute of Education, University of London. 2004.

37.Flegal KM, Carroll MD, Ogden CL, et al. Prevalence and trends in obesity among US adults, 1999-2000. JAMA. 2002;288(14):1723- 1727.

38.Wang G, Dietz WH. Economic burden of obesity in youths aged 6 to 17 years: 1979- 1999. Pediatrics. 2002;109(5):E81-l.

39.Grunbaum J, Kann L, Kinchen SA, et al. Youth Risk Behavior Surveillance — United States, 2001. MMWR. 2002;51(SS04):l-64.

40.Mokdad AH, Ford ES, Bowman BA, et al. Prevalence of obesity, diabetes, and obesity- related health risk factors, 2001. JAMA. 2003;289(l):76-79.

41.Wardle J, Waller J, Jarvis MJ. Sex differ- ences in the association of socioeconomic status with obesity. Am J Public Health. 2002;92(8): 1299-1304.

42.Dykes J, Brunner EJ, Martikainen PT, et al. Socioeconomic gradient in body size and obe- sity among women: the role of dietary re- straint, disinhibition and hunger in the Whitehall II study. Int J Obes Relat Metab Disord. 2004;28(2):262-268.

43.Wardle J, Robb KA, Johnson F, et al. Socio- economic variation in attitudes to eating and weight in female adolescents. Health Psychol 2004;23(3):275-282.

44.Gortmaker SL, Must A, Perrin JM, et al. Social and economic consequences of overweight in

372

Nelson et al

adolescence and young adulthood. N Engl J Med. 1993;329(14): 1008-1012.

45.Bell AC, Adair LS, Popkin BM. Ethnic differ- ences in the association between body mass index and hypertension. Am J Epidemiol. 2002;155(4).346-353.

46.Hu FB, Wang B, Chen C, et al. Body mass index and cardiovascular risk factors in a rural Chinese population. Am J Epidemiol. 2000;151(l):88-97.

47.Pan WH, Flegal KM, Chang HY, et al. Body mass index and obesity-related metabolic dis- orders in Taiwanese and US whites and blacks: implications for definitions of over- weight and obesity for Asians. Am J Clin Nutr. 2004;79(l):31-39.

48.Kann L, Kinchen SA, Williams BI, et al. Youth Risk Behavior Surveillance—United States MMWR. 2000;49(SS05):l-96.

49.Centers for Disease Control and Prevention. Behavioral Risk Factor Surveillance System Survey 1993 (on-line). Available at: http:// www.cdc.gov/brfss/. Accessed June 8, 2005.

50. Centers for Disease Control and Prevention.

Behavioral Risk Factor Surveillance System Survey 1999 (on-line). Available at: http:// www.cdc.gov/brfss/. Accessed June 8, 2005.

51.Dietz WH Jr., Gortmaker SL. Do we fatten our children at the television set? Obesity and television viewing in children and adoles- cents. Pediatrics. 1985;75(5):807-812.

52.Gortmaker SL, Must A, Sobol AM, et al. Television viewing as a cause of increasing obesity among children in the United States, 1986-1990. Arch Pediatr Adolesc Med. 1996;150(4):356-362.

53.Bowman SA, Gortmaker SL, Ebbeling CB, et al. Effects of fast-food consumption on energy intake and diet quality among children in a national household survey. Pediatrics. 2004;113(l Pt 1):112-118.

54.Brener ND, McManus T, Galuska DA, et al. Reliability and validity of self-reported height and weight among high school students. J Adolesc Health. 2003;32(4):281-287.

55.Amett JJ. Emerging adulthood. A theory of development from the late teens through the twenties. Am Psychol. 2000;55(5):469-480.

Am J Health Behav.™ 2007;31(4):363-373 373