Recently, sedentary behaviour among adults and children has received attention as a public health concern(Reference Pate, O'Neill and Lobelo1–Reference Healy and Owen5). Sedentary behaviour is defined as behaviour that does not increase metabolic rate higher than resting and includes sitting(Reference Pate, O'Neill and Lobelo1, Reference Owen, Healy and Matthews4). Sedentary behaviour can be objectively measured through devices worn that measure posture or intensity of activity as well as by self- or proxy-reported time in sedentary behaviours such as sitting for work or transportation, watching television (TV) or playing video games, and computer use for recreation and school work. TV viewing is the most predominant leisure-time sedentary behaviour in children(Reference Sisson, Church and Martin6–Reference Sisson, Broyles and Baker10) and adults(11, Reference Clark, Sugiyama and Healy12). Higher volumes of sedentary time have been associated with mortality(Reference Katzmarzyk, Church and Craig13, Reference Dunstan, Barr and Healy14), cardiometabolic disease(Reference Healy, Dunstan and Salmon15–Reference Hu, Leitzmann and Stampfer17) and obesity(Reference Hu, Li and Colditz18, Reference Jakes, Day and Khaw19) in adults and elevated disease risk(Reference Sardinha, Andersen and Anderssen20, Reference Martinez-Gomez, Eisenmann and Gomez-Martinez21) and overweight and obesity(Reference Rey-Lopez, Vicente-Rodriguez and Biosca22–Reference Crespo, Smit and Troiano24) in children. Several hypotheses have been proposed to explain the relationship observed between TV viewing and health and obesity status. These hypotheses include that: (i) TV viewing displaces time for physical activity; (ii) resting energy expenditure is lower while watching TV; (iii) TV advertising may prompt viewers to consume high-energy snacks during viewing as well as prompts to consume advertised foods later; and (iv) individuals passively overconsume food while watching TV since they are distracted and ignoring cues of satiety.
Evidence from epidemiological studies(Reference Sisson, Broyles and Baker10, Reference Eisenmann, Bartee and Smith25), a time-use study of the after-school period(Reference Atkin, Gorely and Biddle26) and a recent meta-analysis(Reference Marshall, Biddle and Gorely23) support the hypothesis that time spent watching TV may be displacing time spent on physical activity since children with high levels of TV viewing concurrently have low levels of physical activity, although this relationship has a small effect(Reference Marshall, Biddle and Gorely23). There is also evidence to support the contrary (i.e. sedentary behaviour does not displace physical activity)(Reference Ekelund, Brage and Froberg27). An epidemiological study reported no correlation between TV viewing and accelerometer-measured moderate-to-vigorous physical activity (r = 0·013, P = 0·58)(Reference Ekelund, Brage and Froberg27). Furthermore, a cross-sectional clinical study using doubly labelled water showed that the relationship between TV viewing and fatness was not mediated by physical activity in children; although children who watched higher amount of TV were significantly less physically active(Reference Jackson, Djafarian and Stewart28). Further, studies supporting the complexity of these behaviours detail that interventions that have sought to reduce TV viewing were unable to demonstrate subsequent increases in physical activity, even though decreases in overall body weight were reported(Reference Robinson29, Reference Epstein, Roemmich and Robinson30). Some research has examined whether individuals experience a lowered metabolic rate while viewing TV(Reference Cooper, Klesges and Debon31, Reference Klesges, Shelton and Klesges32). However, it is generally accepted that metabolic rate is no lower than in any other sedentary pursuit(Reference Swinburn and Shelly33). There is evidence that indicates advertisements seen by children while watching TV can impact the brands and types of foods they request(Reference Aktas Arnas34–Reference Chamberlain, Wang and Robinson36). Moreover, those children who report paying attention to TV commercials had higher odds of requesting (OR = 3·43; 95 % CI 2·55, 460), buying (OR = 2·73; 95 % CI 2·17, 3·43) and eating snacks while watching TV (OR = 1·60; 95 % CI 1·23, 2·07)(Reference Parvanta, Brown and Du37). Further, cues of satiety have been shown to be disregarded by children(Reference Temple, Giacomelli and Kent38) and college students(Reference Blass, Anderson and Kirkorian39) while watching TV.
Several studies, detailed in a recent review(Reference Pearson and Biddle40), have examined eating patterns and sedentary behaviours in children and adults. Dietary patterns assessed include snacking, fruit and vegetable consumption, fat intake and total energy intake. While many observational studies support the hypothesis that higher sedentary behaviour is associated with poor dietary quality in children and adults, albeit not necessarily total energy intake, not all studies are in agreement and results vary by age(Reference Pearson and Biddle40). While it is very important to understand the relationship between dietary patterns and TV viewing, excess energy intake has been reported to be a predominant driver of the obesity epidemic(Reference Swinburn, Sacks and Ravussin41, Reference Swinburn, Sacks and Lo42). Eight of eleven (73 %) studies in children aged <12 years indicate that higher sedentary behaviour is associated with higher energy intake. Among children aged 12–18 years, three of nine studies reported a positive association. In adults, five of seven studies (71 %) showed a positive association between sedentary time and energy intake. Total energy intake, which is expected to be directly associated with overweight and obesity status, has been studied less frequently than fruit, vegetable, sweetened beverage and snack food intake. Further investigation into the complex relationship between TV viewing and energy intake is warranted to accurately understand the association of these behaviours and obesity, particularly due to the evidence that supports increased energy intake as a primary contributor to global obesity issues. This is particularly relevant in the USA where 34 % of children and 66 % of adults are overweight or obese(Reference Ogden, Carroll and Curtin43) and health costs associated with treatment of obesity and its co-morbidities continue to rise(Reference Thorpe44). The purpose of the present epidemiological study was to conduct an analysis of data obtained from US children and adults to examine the association between energy intake and TV viewing.
Experimental methods
The present study is a cross-sectional, epidemiological analysis of children and adults using data collected as part of the National Health and Nutrition Examination Survey (NHANES) 2003–2004 and 2005–2006.
Subjects
NHANES uses a complex, multistage, probability sampling design to select participants who are representative of the US population. The National Center for Health Statistics ethics review board approved the original survey protocols, and informed consent was obtained for all NHANES participants. Data were obtained for adults and children >2 years of age who participated in both dietary recall interviews. NHANES participants with missing data on gender, ethnicity (Mexican American, European American, African American, other), height, weight, TV viewing and physical activity were excluded from the present analyses. Height and weight were used to calculate the BMI (kg/m2) of each participant. According to BMI, adults were then classified as underweight (<18·5 kg/m2), normal weight (18·5–24·9 kg/m2), overweight (25·0–29·9 kg/m2) or obese (≥30·0 kg/m2)(45). For children, BMI percentile based on age and sex was calculated and classifications were as underweight (<5th percentile), normal weight (5th–84th percentile), overweight (85th–94th percentile) and obese (≥95th percentile)(Reference Kuczmarski and Flegal46). Analyses were stratified by age and conducted separately for pre-school children 2–5 years of age, school-age children 6–11 years of age, adolescents 12–18 years old and adults ≥19 years of age.
Dietary measures
Data on the outcome of interest, total energy intake (kJ/d), were obtained from two 24 h dietary recall interviews conducted by trained technicians. For all analyses, total energy intake from day 1 and day 2 were averaged, a method previously used with national dietary analyses(Reference Bowman47). However, each recall day was examined individually in association with TV viewing to determine if there was a difference. A detailed description of the dietary interviews can be found elsewhere(48, 49); however, procedures are summarized here. The first 24 h recall interview was in-person and took place in the mobile examination centre (MEC). The second 24 h recall interview was conducted via telephone three to ten days following the MEC interview on a different day of the week. Serving sizes and food models were present in the MEC for the initial interview. Participants were given a booklet with information on serving size for the follow-up telephone interview. For children aged 2–5 years, interviews were conducted with the proxy adult who was most knowledgeable about the child's dietary intake, likely a parent. For children aged 6–11 years, interviews were conducted with the child with adult proxy assistance. The US Department of Agriculture's (USDA) Automated Multiple Pass Method, a computerized system, guided the interviewers to prove and collect easily forgotten foods and beverages.
Participants were asked about food consumed in the previous 24 h. In addition, they were asked to compare the day recalled to usual dietary intake (i.e. more than usual, less than usual or usual intake). For pre-school children, 81·9 % and 88·1 % of the recalled days were usual for day 1 and day 2 of the dietary recall interview. For school-aged children, 74·2 % and 88·2 % of the recalled days were usual. For adolescents, 67·2 % and 79·1 % of day 1 and day 2 interviews, respectively, were reported as usual dietary intake. For adults, 73·9 % and 79·0 % of day 1 and day 2 interviews, respectively, were reported as usual intakes.
The USDA Food and Nutrient Database for Dietary Studies (FNDSS), version 2·0 (for 2003–2004) and version 3·0 (for 2005–2006; USDA Agricultural Research Service, Beltsville, MD, USA), was used for processing the reported dietary intakes. The FNDDS provides comprehensive information that can be used to code the individual foods and portion sizes reported by participants, and also nutrient values for calculating nutrient intakes.
Television viewing and physical activity measures
TV and physical activity measures were collected through an interview conducted in the participants’ home (ages 2–11 years and ≥16 years) or the MEC (12–15 years). For children aged 2–11 years, a proxy completed the interview questions.
To assess TV viewing, participants were asked ‘Over the past 30 d, on average about how many hours per day did you sit and watch TV or videos?’ Possible responses included less than 1 h, 1 h, 2 h, 3 h, 4 h and 5 h or more, and were collapsed into low (≤1 h/d), middle (2–3 h/d), and high (≥4 h/d).
Physical activity for participants 2–11 years of age was assessed using the single question, ‘How many times per week does (child) play or exercise enough to make (him/her) sweat and breathe hard?’ Responses were reported in whole numbers from 0 (never) to 77. For participants ≥12 years of age, minutes of moderate-to-vigorous physical activity was assessed through a series of questions. These questions asked participants to report whether or not they had participated in vigorous or moderate physical activity for at least 10 min over the past 30 d. If participants responded ‘yes’ to engaging in moderate or vigorous physical activity, information on the specific activities and the duration was collected. If participants responded ‘no’ moderate or vigorous physical activity, daily active time was recorded as zero. If participants responded ‘yes’, then the individual time spent in each moderate and vigorous activity was summed to provide a total time in activity per day.
Data analysis
The SAS statistical software package version 9·2© (SAS Institute, Cary, NC, USA) was used for all analyses. Descriptive statistics including means, standard errors and percentages were calculated to summarize population characteristics. Descriptive information has been presented in kJ; however, outcome analyses were conducted using kcal to maintain manageable parameter estimates. Preliminary analyses revealed an interaction between sex and TV viewing and therefore analyses were stratified by sex. Mean daily total energy intake was calculated using least-squared means for each age and sex group adjusted for age in years, BMI (or percentile for children), physical activity and ethnicity. The relationship between TV viewing and daily energy intake was examined by a series of multivariate linear regression models. The first model included TV viewing, BMI (percentile for children aged 2–18 years), age in years and ethnicity. The second model included all variables for model 1 with the addition of either minutes of moderate-to-vigorous physical activity (adolescents and adults) or times per week of hard play (pre-school and school-age children). Appropriate survey analyses and weight variables were used to ensure national representation of the findings.
Results
Data were examined for outliers in dietary and physical activity measures. A conservative and systematic measure to eliminate outliers in addition to inaccurate data as a result from memory or over- and under-reporting issues(Reference Thompson and Byers50) was employed and data from participants whose total energy intake fell within the highest and lowest 2 % of the distribution were not included in these analyses. Data were excluded for pre-school children if total energy intake was less than 3424·2 kJ/d (n 38) or if it exceeded 12 884·5 kJ/d (n 16). For school-aged children, participants were removed if total energy intake was less than 4123·2 kJ/d (n 33) or if it exceeded 13 755·2 kJ/d (n 48). For adolescents, participants were removed if total energy intake was less than 3415·8 kJ/d (n 61) or exceeded 19 000·3 kJ/d (n 49). For adults, participants were removed if total energy intake was less than 3466·0 kJ/d (n 112) or exceeded 18 066·8 kJ/d (n 260). Additionally, if daily moderate-to-vigorous physical activity exceeded 12 h, those adolescents (n 8) and adults (n 4) were removed (physical activity was assessed differently in younger children). Final analytical sample sizes for the pre-school children, school-aged children, adolescents and adults were n 1369, n 1668, n 3233, and n 7850, respectively.
Descriptive characteristics of each sample can be found in Table 1. The mean participant age was 3·5, 8·4, 15·0 and 46·1 years for pre-school children, school-age children, adolescents and adults, respectively. Females consistently reported lower daily energy intake. In females, the pre-school girls reported 6580·8, school-aged girls reported 7841·5, adolescent girls reported 7925·6 and women reported 7266·4 kJ/d. In males, the pre-school boys reported 6908·0, school-aged boys reported 8707·0, adolescent boys reported 10 089·6 and men reported 9782·1 kJ/d. In both males and females, the 12–18 years age group reported the highest daily level of energy intake. The percentage of the sample in the high TV viewing category was 13·7 % for pre-school children, 17·2 % for school-aged children, 20·3 % for adolescents and 21·2 % for adults (P < 0·0001).
Descriptive characteristics of young children (2–11 years), older children (12–18 years) and adults in analyses examining TV viewing and energy intake, 2003–2006 National Health and Nutrition Examination Survey
TV, television; MVPA, moderate-to-vigorous physical activity; n/a, not applicable.
*For adults classifications are underweight <18·5 kg/m2, normal weight 18·5–24·9 kg/m2, overweight 25·0–29·9 kg/m2 and obese ≥30·0 kg/m2. For children obesity status is based on BMI percentiles for age and sex with underweight <5th percentile, normal weight 5th–84th percentile, overweight 85th–94th percentile and obese ≥95th percentile.
†Different questions were asked to children aged 12–18 years and adults v. children aged 2–11 years. Adolescents and adults were asked about minutes of MVPA in the previous month, which was used to calculate weekly minutes of MVPA. Pre-school and school-aged children's parent proxy was asked about the times per week the child engaged in exercise or play enough to sweat and breathe hard.
Total energy intake by TV viewing category among age and sex categories can be found in Table 2. There was no significant association between TV viewing category and energy intake for any of the age and sex groups, with the exception of the adolescent girls and men (Table 3). These findings were similar for analyses with and without physical activity for both adolescent girls and men. In the adolescent girls, the high TV viewing category consumed more energy daily (816·3 kJ (195 kcal) and 820·5 kJ (196 kcal)) than the low TV viewing category (P = 0·03) for models with and without physical activity, respectively. Men in both the middle and high TV viewing categories consumed less energy daily than men in the low TV viewing category. For both models with and without physical activity, men in the middle TV viewing category consumed 548·4 kJ (131 kcal) less (P = 0·0002); while men in the high TV viewing category consumed 473·0 kJ (113 kcal) (P = 0·02) and 464·7 kJ (111 kcal) less (P = 0·02) than men in the low TV viewing category for models with and without physical activity, respectively. These findings are in the direction opposite of what would be expected (i.e. higher energy intake in those with higher TV viewing).
Total adjusted energy intake (kJ/d) for category of TV viewing by age and gender groups, 2003–2006 National Health and Nutrition Examination Survey
TV, television.
Least-squared means adjusted for age, BMI (percentile for all children), physical activity (daily minutes of moderate-to-vigorous physical activity (MVPA) for children aged 12–18 years and adults and weekly times of hard play for children aged 2–11 years), ethnicity and smoking status (adults only).
Different questions were asked to children aged 12–18 years and adults v. children aged 2–11 years. Older children and adults were asked about minutes of MVPA in the previous month, which was used to calculate weekly minutes of MVPA. Pre-school and school-aged children's parent proxy was asked about the times per week the child engaged in exercise or play enough to sweat and breathe hard.
Relationship between TV viewing and energy intake by age and gender groups, 2003–2006 National Health and Nutrition Examination Survey
TV, television; Ref., referent category.
Different questions were asked to children aged 12–18 years and adults v. children aged 2–11 years. Adolescents and adults were asked about minutes of moderate-to-vigorous physical activity (MVPA) in the previous month, which was used to calculate weekly minutes of MVPA. Pre-school and school-aged children's parent proxy was asked about the times per week the child engaged in exercise or play enough to sweat and breathe hard.
The model without physical activity was adjusted for BMI (percentile for all children), age and ethnicity. The model with physical activity included all previously listed covariates and either minutes of MVPA (children 12–18 years and adults) or times per week of hard play (children aged 2–11 years).
Analyses were conducted with average total kcal rather than kJ to prevent extremely large parameter estimates.
Bold indicates significance (P < 0·05).
Discussion
The present study was unable to show an association between TV viewing and total energy intake in US pre-school children, school-aged children, adolescent boys and adult women. However, an association between TV viewing and total energy intake (kcal/d) was observed among girls aged 12–18 years and men. Adolescent girls in the highest TV viewing category (≥4 h/d) were estimated to consume 820·5 kJ (196 kcal) more energy daily than adolescent girls in the lowest TV viewing category (≤1 h/d). These findings are similar to those reported in previous studies that report total energy intake (kcal/d)(Reference Crespo, Smit and Troiano24, Reference Jeffery and French51–Reference Rosenberg, Norman and Sallis53) but conflict with others(Reference Bowman47, Reference Miller, Taveras and Rifas-Shiman54–Reference Gore, Foster and DiLillo57). Men in the middle (2–3 h/d) and high (≥4 h/d) TV viewing categories were estimated to consume 464·7–473·0 kJ (111–131 kcal) less energy per day compared with men in the lowest TV viewing category (≤1 h/d). This is in contrast with previous reports of a positive(Reference Bowman47) or null(Reference Jeffery and French51) relationship. It is worth noting that all age and sex groups reported dietary intake within the recommended ranges of 6694–10 042 kJ (1600–2400 kcal) for women and 8368–12 552 kJ (2000–3000 kcal) for men; albeit energy intake needs for weight maintenance depend on the age, sex, size and physical activity of the individual(Reference VanHorn, Fukagawa and Achterberg58). Dietary intake and food consumption patterns have been previously reported in a number of studies and are summarized in a review published recently(Reference Pearson and Biddle40). This review states that while the majority of studies show a positive association between TV viewing and diet, not all studies agree(Reference Pearson and Biddle40). These discrepancies can likely be attributed to differences in methodology of dietary quality or intake, characteristics of the sample and the complicated nature of the relationship between diet, TV viewing and adiposity.
Measurements of dietary quality have been defined through various methods including but not limited to amount of snacking, fruit, vegetable, fast food, sugar-sweetened beverage consumption, total energy intake, FFQ, behavioural questionnaires and 24 h recalls. The current study utilizes data obtained from two 24 h dietary recalls, taken 3 to 10 d apart by a trained interviewer using the automated multiple-pass method to calculate an average of total energy intake. For children under the age of 11 years, proxy respondents assisted or completed the interview on behalf of the child. This methodology was used in a previous study by Bowman(Reference Bowman47) using the USDA's Continuing Survey of Food Intakes by Individuals 1994–1996, a nationally representative sample of Americans. Bowman observed that TV viewing was positively associated with energy intake. Men and women who watched >2 h TV/d consumed 573·5 kJ (137 kcal) more energy daily than adults who watched <1 h/d(Reference Bowman47). Bowman's results, with nearly identical methodology of data collection, are incongruent with the current findings. Differences may be attributed to differences in analyses; Bowman did not stratify by sex while our analysis does.
In a study that included children and examined the association between total energy intake using similar methodology of dietary recall, Rosenberg et al.(Reference Rosenberg, Norman and Sallis53) found no association between TV viewing and energy intake. Congruent with our findings, Crespo et al.(Reference Crespo, Smit and Troiano24) reported an association only in girls such that those viewing >5 h TV/d consumed 732·6 kJ (175 kcal) more energy daily compared with those girls viewing TV for ≤1 h/d. Studies that used FFQ to measure total energy intake consistently showed increasing hours of TV viewing was associated with higher total energy intake (205·1–699·1 kJ/d (49–167 kcal/d)) in children ranging from 3 to 12 years of age(Reference Miller, Taveras and Rifas-Shiman54–Reference Wiecha, Peterson and Ludwig56). Energy intake calculated from food frequencies could be different from that in a food recall since kJ/d are determined based on reported patterns of food intake in the food frequency rather than precisely what was eaten on the day prior to the recall. It is noteworthy that two of those studies noting an increase in energy intake (as measured by food frequency) and TV viewing(Reference Sonneville and Gortmaker55, Reference Wiecha, Peterson and Ludwig56) were from the same Planet Health Study and thereby the same sample of children. Interestingly, a single study using five consecutive days of a weighed food record showed no associated between energy intake and TV viewing, although it was only in sixty adolescents(Reference Ortega, Andres and Requejo52).
The present study was only able to demonstrate a positive association between total energy intake (kcal/d) and TV viewing among adolescent girls. The reason why adolescent girls may be at greater risk for higher energy intake with TV viewing is not entirely clear. Boys have consistently been reported to engage in higher volumes of screen-based activities(Reference Sisson, Church and Martin6–Reference Gorely, Marshall and Biddle8) and a higher proportion of boys were found to exceed the American Academy of Pediatrics’ TV viewing recommendation of <2 h/d(Reference Sisson, Broyles and Baker59). Pearson and Biddle's(Reference Pearson and Biddle40) review did conclude that there is a trend towards more consistent relationships between females regarding TV viewing and dietary quality than in males. Differences in social and occupational roles between men and women have also been proposed as reasons to explain this relationship observed in women such that the interaction between TV viewing, fast-food eating and weight control are different between genders(Reference Jeffery and French51).
Additionally, the results of the present study suggest a negative association between TV viewing and energy intake in men. As noted in the review by Pearson and Biddle(Reference Pearson and Biddle40), no previous studies in men have reported a negative association between TV viewing and energy intake, although two reported no association(Reference Jeffery and French51, Reference Gore, Foster and DiLillo57). A laboratory-based study found that adult participants consumed more food when given a meal while watching TV(Reference Bellisle, Dalix and Slama60). A study conducted in children 12–16 years of age that decreased sedentary behaviour by 25–50 % for three weeks reported a decrease in energy intake during that time(Reference Epstein, Roemmich and Paluch61). However, when the reverse experiment was performed (increased sedentary behaviours by 25–50 % for three weeks) there was not a concomitant increase in energy intake(Reference Epstein, Roemmich and Paluch61). The interventions were counterbalanced in the order received by participants to prevent any timing effect. It is clear that the relationship between TV viewing and dietary patterns is complex. When total energy intake, which would most directly be associated with overweight and obesity, is the dietary outcome of interest, the findings are mixed.
The findings of the current study indicate TV viewing is not positively associated with energy intake, except in 12–18-year-old girls. Additionally, TV viewing is negatively associated with energy intake in men. These findings should be interpreted in light of the study's strengths and weaknesses. The study had a large sample size and utilized data obtained from a nationally representative sample of US men, women and children. Additionally, both 24 h dietary recalls were administered by highly trained NHANES interviewers with the assistance of the automated multiple-pass method. These data allowed us to estimate the average total energy (kcal/d) intake for all subjects. Furthermore, participants were sent home with food models and portions size aids to enhance the accuracy of the telephone recall interview. However, limitations in the accuracy, honesty and social desirability of intake in dietary recall methodology are always a concern(Reference Hebert, Hurley and Peterson62–Reference Miller, Abdel-Maksoud and Crane64). It is also worth noting that different interview methodologies (i.e. in-person v. telephone) were used for the day 1 recall and the day 2 recall. There were no major differences in the association between TV viewing and energy intake between day 1, day 2 and the average with the exception of adolescent girls. The day 2 and averaged analyses were similar while day 1 showed a similar pattern; however statistical significance was not reached. Usage of proxy report and assistance is an additional limitation since proxies may not know exactly what was consumed on the day prior to the interview. TV viewing is also self- or proxy-reported in a categorical fashion. Observation of TV viewing or electronic TV monitors is a more accurate method to measure TV viewing. Additionally, reporting the continuous number of hours of TV viewing would allow greater flexibility and precision in analysis.
Conclusions
In summary, TV viewing in adult women and children in the 2003–2006 NHANES was not associated with energy intake, with the exception of adolescent girls and adult men. Previous evidence generally shows that higher TV viewing is associated with poorer dietary quality and higher energy intake, although not all studies support this relationship(Reference Pearson and Biddle40). It is recommended that future studies explore the relationship of TV viewing and energy intake in children and adults due to the discrepant research findings. This is particularly relevant to understand the differences between males and females regarding TV viewing and energy intake. It is likely that TV viewing is associated with diet(Reference Pearson and Biddle40); however, it may be due to dietary patterns, such as fruit, vegetable and fast-food consumption, than energy intake per se. Another measurement issue future studies may explore would be to ascertain the precision with which people can recall the quantity of food they consume while viewing TV. A recent opinion article addresses the hypothesis that eating while viewing TV leads to distraction(Reference de Graaf and Kok65) which could potentially impact energy intake as well as recall of food consumed. Viewing TV while eating may lead to under-reporting of energy intake if portion sizes are underestimated or not recalled. There is some evidence, at least in adults(Reference Oldham-Cooper, Hardman and Nicoll66, Reference Higgs and Woodward67), that watching TV while eating impairs memory of food consumed during the meal. Therefore, examination of type of foods consumed such as fast food, energy-dense foods or fruit and vegetables may provide better explanations of food patterns v. 24 h recalls. This may be particularly true because of social desirability leading to under-reporting of intake, compounded with inability to accurately recall the quantity of food consumed while TV viewing, leading to further under-reporting. New technologies that can be used with food reporting such as digital photography pre and post meal could be another avenue to more accurately explore the relationship of energy intake and TV viewing. The dearth of research studying this phenomenon, especially in children, warrants further studies. Other areas for future research should include examination of the hypothesis related to precisely how and why TV viewing is associated with overweight and obesity. The complexity of these relationships is clear and well-designed studies are needed to gain deeper understanding and evidence for or against each hypothesis. Future research should also examine longitudinal and interventions where TV viewing is manipulated to discern if this relationship is amenable to change.
Acknowledgements
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. There are no conflicts of interest. S.B.S. and S.T.B. designed the research; S.B.S., L.B. and C.R. analysed the data; S.B.S., S.T.B., L.B., C.R. and M.L. wrote the paper; S.B.S. had primary responsibility for final content. All authors read and approved the final manuscript.
