Television Watching, Energy Intake, and Obesity in US Children: Title and subTitle BreakResults From the Third National Health and Nutrition Examination Survey, 1988-1994 FREE

MORE US adults are overweight now than 20 years ago and the prevalence of overweight among American children and adolescents is also increasing.^{1 - 5} Troiano et al³ reported on the trends of overweight status using data from several national cross-sectional surveys of US children. Overweight was defined by age- and sex-specific 85th and 95th percentiles of body mass index (BMI) using data from cycles 2 (1963-1965) and 3 (1966-1970) of the National Health Examination Survey (NHES). Using data from phase 1 (1988-1991) of the Third National Health and Nutrition Examination Survey (NHANES III), they reported that the prevalence of overweight was 22% when defined by the 85th percentile and 11% when defined by the 95th percentile. A subsequent update using the data from both phases of NHANES III (1988-1994) suggested that the increase in overweight prevalence had continued between 1991 and 1994.⁶

It is clear that obesity is detrimental to the health of adults, and an increasing number of studies show that overweight children are at higher risk of developing dyslipidemia, hypertension, diabetes mellitus, and other weight-related morbidities.^{7 - 12} Behavioral, environmental, and social factors in addition to a genetic predisposition may be associated with a child becoming overweight. Other factors associated with childhood obesity include time spent in front of a television or computer monitor, availability of facilities for participation in physical activities, and the opportunities to safely engage in sports, play, and exercise.^{13 - 20}

Since children are in a continuous state of physical growth, their energy needs in relation to body weight are difficult to study in large epidemiologic studies. Changes in energy intake in 2 large population-based studies show that even though the reported total energy and fat intake has not increased substantially in England or the United States in the last decade, the number of overweight persons has increased in both countries.^{1 ,21 - 22} Neither the contribution of declining physical activity or increasing sedentary behavior to the apparent positive energy balance^{23 - 25} nor the association between energy intake and sedentary behaviors using a national representative sample of US children has been established.

In this article, we investigate the association between obesity, sedentary behavior, and energy intake among children and adolescents. The purpose of this study is to examine how the prevalence of obesity is associated with physical activity, television watching, and energy intake using a national representative sample of US children aged 8 through 16 years.

The NHANES III was conducted by the National Centers for Health Statistics of the Centers for Disease Control and Prevention, Atlanta, Ga. This survey, as in previous NHES and NHANES, was designed to provide national reference data for a variety of diseases, chronic conditions, and health indicators of the population of the United States and selected subpopulations. A detailed explanation of NHANES III has been previously described.²⁶ Briefly, NHANES III is a nationally representative sample of the civilian, noninstitutionalized population of the United States. It oversampled the very young and the very old, blacks, and persons of Mexican ancestry to produce reliable estimates of their health conditions.

The NHANES III consisted of a home interview and a detailed clinical examination conducted in a mobile examination center. Our analytic sample consists of 4069 children, aged 8 to 16 years, who completed the physical activity questionnaire and the body measurement component at the mobile examination center, and who also responded to a home interview. Those children who had a proxy answer to the physical activity and television questions in the mobile examination center were excluded from our analyses (N = 9). Interviewing staff consisted of experienced persons, many of whom were of Hispanic origin and/or were bilingual in English and Spanish. All staff attended yearly training sessions to ensure maintenance of effective interviewing and body measurements skills. Information on the respondent's race and ethnicity was used to classify persons as non-Hispanic white, non-Hispanic black, or Mexican American. Age was defined as the age in years at the time of the household interview.

Participating children were asked: "How many times per week do you play or exercise enough to make you sweat or breathe hard?" These activities did not exclude school-related involvements such as physical education. The interview also included a question on the number of hours of television watched the previous day. We estimated the percentage of children who reported being active most days of the week as children who reported participating in physical activity at least 5 times per week as recommended by the Surgeon General's Report on Physical Activity and Health.^{27 - 28} Participants were classified into 7 categories of physical activity from 1 or less time per week to 8 or more times per week.

Assessment of hours of television watched the previous day was obtained during the home interview and again during the mobile examination center visit. These 2 interviews were approximately 1 to 3 weeks apart. Television watching may vary from day to day (2-day agreement: κ = 0.10). We therefore used the average of the 2-day television-watching recall to estimate the percentage of children watching television for 1 hour or less, 2 hours, 3 hours, 4 hours, and 5 or more hours per day.

Body composition was estimated using BMI, which is calculated as weight in kilograms divided by the square of height in meters, since this is significantly and positively related to percentage of body fat and total body fat in boys and girls.^{29 - 36} We used data from cycles 2 and 3 from the NHES to establish age- and sex-specific cutoff points for the definition of obesity. Cycle 2 was conducted from 1963 to 1965 and examined children aged 6 to 11 years, while cycle 3 was conducted from 1966 to 1970, and examined adolescents between the ages of 12 and 17 years. These 2 cycles of NHES provide the earliest national data for height and weight between children and adolescents. We defined obesity as the 95th percentile of BMI from data obtained during the NHES (1963-1970), which has a high specificity for excess body fat in children.^{3 - 4}

As part of the examination, a trained diet technician obtained information on food intake using a 24-hour dietary recall. The dietary interviewers used the Dietary Data Collection system, which is an automated standardized interactive dietary interview and coding system, and was specifically designed for NHANES III by the Minnesota Nutrition Coordinating Center. Participants were asked to report all foods and beverages consumed, excluding plain drinking water, during the previous 24 hours, from midnight to midnight. The food database for this system was linked to the US Department of Agriculture's (USDA) Survey Nutrition Database and produced total energy intake.²⁶

Statistical analyses were carried out using SAS, SUDAAN, and STATA.^{36 - 38} For NHANES III, sampling weights were calculated taking into account the unequal probability of selection resulting from the cluster design and from planned oversampling of certain subgroups. All analyses incorporated the sampling weights. We used the svymean estimation procedure in the STATA software to calculate the prevalence, SEs and 95% confidence intervals, and the PROC RLOGIST from SUDAAN to calculate estimated relative risk and the 95% confidence intervals.

Table 1 presents the cross-sectional prevalence estimates of hours of television watched per day by sex, age, and race/ethnicity among US children. Overall, almost half of US children aged 8 to 16 years watched more than 2 hours of television a day. More boys than girls watched more than 2 hours of television (38% and 48%, respectively). Among race/ethnicity groups, a higher percentage of non-Hispanic black (65%) and Mexican American (53%) children watched television for 3 or more hours than non-Hispanic white children (37%). On the average, 17% of non-Hispanic black children watched 5 hours or more a day, whereas only 9% of Mexican American and 6% of non-Hispanic white children watched television for 5 or more hours a day.

Table 2 shows the prevalence of obesity, total energy intake, and percentage of children who are active 5 or more days per week. Total energy intake was consistently higher among boys than girls, regardless of age or race/ethnic groups. Boys reported being more physically active than girls; however, only 56.7% of US children aged 8 to 16 years engage in physical activity 5 or more days per week. Non-Hispanic white boys had the highest prevalence of participating in physical activity 5 or more days per week (72.4%), while non-Hispanic black girls had the lowest prevalence of participating in physical activity 5 or more days per week (41.8%).

Figure 1 shows the prevalence of obesity among US children according to hours of television watched per day. The prevalence of obesity increased as hours of television watching increased. Obesity was lowest among children who reported watching 1 hour or less of television per day. The prevalence of obesity according to weekly participation in vigorous physical activity showed no clear trends for either boys or girls (data not shown).

We subsequently examined whether television watching was related to energy intake in boys and girls. Figure 2 shows adjusted energy intake by hours of television watching in boys and girls. After adjusting for age, BMI, race/ethnicity, family income, and weekly bouts of physical activity, we found that total energy intake was positively associated with hours of television watched more so among girls (R = 0.43) than among boys (R = 0.26). The mean adjusted energy intake of girls who watched 1 hour or less of television was 7748 kJ/d, whereas mean energy intake of girls who watched 5 or more hours of television a day was 8468 kJ/d. These results indicate that energy intake has a tendency to increase with increased television watching, especially when comparing those watching 1 hour or less of television and those watching 5 or more hours a day.

Table 3 shows the estimated relative risk of obesity by hours of television watching after adjusting for several correlates (eg, age, race/ethnicity, family income, energy intake, physical activity). A higher prevalence of obesity was significantly associated with higher television watching among girls. We found no consistent association between prevalence of obesity and weekly bouts of vigorous physical activity.

Our findings provide estimates of the distribution of television-watching habits in US children aged 8 to 16 years and how television watching relates to obesity. We found that increased television watching is associated with a higher prevalence of obesity among girls, but not among boys. Children who watched the most number of hours of television a day had the highest prevalence of obesity; this held true after controlling for age, race/ethnicity, and family income (Figure 1) (Table 3).

Boys (12.4%) exhibit a somewhat higher prevalence of obesity than girls (10.9%). More alarming is the fact that the prevalence of obesity has more than doubled in less than 1 generation. Using the age- and sex-specific 95th percentile of BMI as the cutoff point from data collected between 1963 and 1970 NHES, our results indicate prevalence estimates of obesity to be greater than 10%. If in 1963 through 1970, the top 5% were classified as obese, we now have approximately 12% of US children in this category.

We found that energy intake has a tendency to increase with increased television watching, with girls consuming on average an extra 732.2 kJ (175 kcal) a day when comparing those watching 1 hour or less of television with those watching 5 or more hours a day. When taking into account that the hours spent watching television represent hours doing little or no physical activity,^{39 - 40} and the cumulative effect of these extra 732.2 kJ (175 kcal) a day, the net result could be a positive energy balance that may be linked to childhood and adult obesity.

Our finding that television watching was more closely associated with obesity than physical activity may be partially explained by the lack of sensitivity of the NHANES III questionnaire to accurately measure physical activity in children. We must, therefore, be cautious in interpreting these results based on one 24-hour dietary recall and 1 question on self-reported weekly bouts of physical activity in this group of children, because of the substantial intraindividual and interindividual variability. Thus, there is a possibility that the associations we have observed with obesity may be attenuated because of the unreliability of these measures.^{41 - 42} In addition, the ability to accurately measure both physical activity and television-watching habits are influenced by 2 factors. First, how well does self-report of physical activity reflect what children and adolescents are actually performing over the time frame that the question is being asked? Second, can a single report of physical activity accurately represent the usual physical activity patterns of children and adolescents? It may be that multiple days of observation during the year, over several years, are needed to reliably estimate the usual physical activity and television-watching patterns of young children.

Boys reported engaging in more physical activity (66.7% are active ≥5 d/wk) than girls (46.1% are active ≥5 d/wk), and they consume more kilojoules (10 506 kJ/d [2511 kcal/d]) than girls (7945 kJ/d [1899 kcal/d]). Boys and girls may be different in how they deal with weight gain. For example, we found more girls than boys, aged 12 to 16 years, reporting that they had attempted to lose weight in the past 12 months (22.6% and 9.9%, respectively). Our findings are similar to those of Steen et al,⁴³ who found that the prevalence of overweight was not substantially different between boys and girls. Other researchers have also found that despite the high prevalence of obesity in boys, few of them reported trying to lose weight.^{43 - 48} The boys perceived themselves as significantly less overweight and were happier with their "looks" than girls. Also, girls seem to be less likely to think of themselves as exercisers. More research is needed to understand how weight loss practices and body image perceptions are related to energy intake, television watching, and physical activity in boys and girls.

Previous studies have found an indirect association between these 2 behaviors (physical activity and television watching). DuRant et al²⁴ assessed television watching and physical activity by direct quantitative observation and its association with body composition in 5- to 6-year-old children. As expected, children were less active during their longest bout of television watching, while physical activity increased in periods of low television watching. Correspondingly, the authors concluded that children engage in more physical activity when they are not watching television. We conducted a regression analysis to examine the relationship between daily hours of television watching and weekly bouts of physical activity. We found an inverse association between television watching and physical activity for both boys (β coefficient, −0.06; P = .02) and girls (β coefficient, −0.075; P = .01). Although small, this correlation has been confirmed in other validity studies among children.⁴⁹

Our study is also limited to television watching and does not include time spent on the computer. More research is needed to develop better tools to assess physical activity and to assess time spent on the computer or playing computer games in addition to just watching television. Another limitation of our study is that our findings are cross-sectional and cannot distinguish if high television watching causes obesity or if being obese causes children to watch too much television. Our findings do confirm that the prevalence of obesity has increased since 1963, and that energy intake in conjunction with excess television watching and little physical activity may play a role.

Our results show that the prevalence of obesity is greater among children who watch 4 or more hours of television a day. Our findings also show that girls report engaging in less physical activity than boys, and provide support for a major national campaign to promote reductions in sedentary behavior, such as television watching, and to increase the opportunities for both boys and girls to participate in lifetime physical activities. Sound dietary practices and reductions in sedentary behavior may help in preventing further increases in the prevalence of obesity of US children.

Accepted for publication October 4, 2000.

Corresponding author: Ross E. Andersen, PhD, Division of Geriatric Medicine and Gerontology, The Johns Hopkins School of Medicine, 4940 Eastern Ave, Suite 025, Baltimore, MD 21224 (e-mail: andersen@jhmi.edu). Reprints: Carlos J. Crespo, DrPH, MS, Department of Social and Preventive Medicine, School of Medicine and Biomedical Sciences, State University of New York at Buffalo, 270 Farber Hall, Buffalo, NY 14214-3000 (e-mail: ccrespo@buffalo.edu).