International Journal of Pediatric Obesity, 2010; 5: 420–427
ORIGINAL ARTICLE
Neighbourhood food environment and dietary intakes in adolescents: Sex and perceived family affluence as moderators
SAI-YIN HO1, BONNY YEE-MAN WONG1, WING-SZE LO1, KWOK-KEI MAK1, G. NEIL THOMAS2 & TAI-HING LAM1 1School
of Public Health, University of Hong Kong, Pokfulam, Hong Kong SAR, 2Department of Public Health and Epidemiology, University of Birmingham, UK
Abstract Objective. To examine the effects of perceived availability of fast-food shops, restaurants, and convenience stores on adolescent dietary intakes. Methods. Survey data from 34 369 students in 42 Hong Kong secondary schools were collected in 2006–7. Respondents reported the availability of fast-food shops, restaurants and convenience stores in the neighbourhood, and their intakes of fruit, vegetables, high-fat foods and junk food/soft drinks. For intakes of high-fat foods and junk food/ soft drinks, ⱕonce a week was defined as low consumption and the rest moderate/high consumption. At least three servings of vegetables and two servings of fruit daily were defined as sufficient consumption. Logistic regression yielded adjusted odds ratios (OR) for each dietary intake in relation to the reported food shops. Potential effect modifications by sociodemographic factors were also examined. Results. Perceived availability of fast-food shops and convenience stores were positively associated with moderate/high consumptions of high-fat foods (ORfast⫽1.10 and ORcon⫽1.15) and junk food/ soft drinks (ORfast⫽1.10 and ORcon⫽1.10). Significant negative associations of the perceived availability of restaurants with intakes of vegetables and fruit were observed (ORveg⫽0.87 and ORfruit⫽0.83). The positive relationship between reporting fast-food shops with intake of junk food/soft drinks were observed only in boys and those with low perceived family affluence. The negative association of reporting restaurants with fruit consumption was found in those with low and middle perceived family affluence only. Conclusions. Perceived availability of neighbourhood fast-food shops, restaurants, and convenience stores may have a negative impact on adolescent dietary intakes particularly for those from poorer families. Key words: Adolescents, environment, neighbourhood, food shops, dietary intakes, vegetables, fruit, high-fat food, soft drinks, and junk food
Introduction Obesity in young people has become an emerging public health problem worldwide (1). Despite the well-established evidence on the association between low fruit and vegetables intake and high intake of food high in fat and/or sugar with higher risk of obesity (2), increased consumption of salty snacks, soft drinks and fried food, and low fruit and vegetable intake in adolescents were observed (3–5). Given the development of eating behaviours in adolescence and the tracking of obesity from adolescence to adulthood (1), the factors that influence adolescent dietary intake warrant detailed investigation. According to social-ecological models of health behaviour (6), eating behaviours are functions of
personal, social, and environmental factors and their interactions. Social-ecological models emphasize the role of distal (e.g., neighbourhood) and proximal (e.g., home) physical environments as deterrents or facilitators of healthy eating behaviours. Some of the commonly examined distal/proximal environmental characteristics, in the context of eating behaviours in youth, include the availability of food shops in the neighbourhood and the availability of foods in the home/school (5,7–13). Several studies have found a positive relationship between the availability of fruit and vegetables in the home with greater fruit and vegetable consumption among adolescents. Likewise positive associations between the availability of unhealthy
Correspondence: Dr. S. Y. Ho, School of Public Health, The University of Hong Kong, 21 Sassoon Road, Pokfulam, Hong Kong. Fax: 852 2855 9528. Email:
[email protected]
(Received 18 May 2009; accepted 9 November 2009) ISSN Print 1747-7166 ISSN Online 1747-7174 © 2010 Informa Healthcare DOI: 10.3109/17477160903505910
Environment and dietary intakes food in home/school and a higher intake of unhealthy food and lower intake of fruit and vegetables were observed (9–10,13). Adolescent dietary intake may also be influenced by the presence of neighbourhood food shops. Eating in fast-food shops or restaurants was associated with a higher intake of fat, sugar-sweetened drinks, processed meat, and lower intake of fruit and vegetables among adolescents (14–18). Some studies have also reported that the presence of neighbourhood food shops were associated with obesity in adults (19–21). It has been argued that the availability of fast-food shops, restaurants and convenience stores in the neighbourhood, one aspect of the obesogenic environment, encourages unhealthy food consumption and discourages healthy food consumption, and hence may lead to obesity (22). However, it is unclear whether the availability of fast-food shops, restaurants and convenience stores alone is sufficient to influence intake. There are socio-economic disparities in the availability of food shops, variety of food choice, nutrient knowledge and consumption of fruit and vegetables (23–26). Individuals with lower socio-economic status and lower education were likely to have less access to healthy choices and poorer nutrient knowledge, and hence they might be more vulnerable to their neighbourhood food environment. However, such data are currently not available. The identification of environmental attributes of dietary intakes in different socio-demographic subgroups can assist in the planning of more effective environmental interventions aimed to prevent obesity by reducing obesogenic eating behaviours in adolescents.We therefore investigated the relationship between the perceived availability of fast-food shops, restaurants, and convenience stores in the neighbourhood with dietary intakes among Hong Kong Chinese adolescents and potential moderation by socio-demographic factors.
Methods Subjects This study was part of the Hong Kong Student Obesity Surveillance (HKSOS) project. Ethical approval was granted by the Institutional Review Board. Schools were randomly sampled with stratification by district, source of funding, language of instruction (English/Chinese), religious background (Catholic/ Christian/Buddhism/None) and co-education (both sexes/single sex) to represent all main-stream, noninternational secondary schools in Hong Kong. All Form 1–7 students (year 7–13 in the US) in 42 selected schools were invited for voluntary participation with passive parental consent. A total of 34 369
421
(83.8% of 41 027) questionnaires were collected in 2006–7. Non-participants (16.2% of 41 027) were absentees and those who refused to participate. After excluding questionnaires with incomplete data (n⫽8 119) and then extreme body mass index (BMI) values (⬍10 or ⬎50 kg/m2) (n⫽1 454), the sample size was 24 796 (72.1% of 34 369). The excluded and included subjects were similar in sex, age, parental education, and perceived family affluence (Cohen’s d⫽0.06 to 0.11 (27); mean⫽0.09). Moreover, our sample was similar to the Hong Kong adolescent population in age (Cohen’s d⫽0.14) and sex (Cohen’s d⫽0.11) (27). The effect sizes of 0.1, 0.3 and 0.5 are considered small, middle and large, respectively (27). The one-month test-retest reliability of the measures used in this study was assessed with 600 students (41% boys; mean age⫽14.57⫾1.66) from two of the participating schools. Measures Dietary intakes. We assessed the intake of four food items: high-fat foods, junk food/soft drinks, fruit, and vegetables. For high-fat foods (defined as deep fried, fried, or fatty foods) and junk food/soft drinks, we asked how frequently the students consumed each food item with eight response options from never to ⬎twice a day. In our Chinese questionnaire “junk food” is put as “snack”, which includes essentially potato chips, sweets, cookies, etc, but excludes fruit and vegetables in Hong Kong culture. For clarity, the term “junk food” is used here. High-fat foods and junk food/soft drinks had good test-retest agreements of 77.0% and 77.8%, respectively. Low consumption was arbitrarily defined as ⱕone day/week, encompassing about 30% of students, and the rest as moderate/high consumption (ⱖtwo days/week). For sensitivity analysis, a different cut-off point of ⱕone to three days/month was used to examine the robustness of the results. For fruit, we asked for the number of servings consumed each day, where one serving roughly equals a medium-sized apple or orange. For vegetables, we asked for the number of normal-sized bowls of cooked vegetables consumed each day, and one bowl equals two servings. Ten options from none to ⬎four servings of fruit or four bowls of vegetables were provided. These two items had test-retest agreements of 82.8% and 75.4%, respectively, which were comparable with other validated food consumption measures (11). Sufficient consumption was defined as ⱖtwo servings of fruit and ⱖthree servings of vegetables daily according to current recommendations (28). The combination of ⱖtwo servings of fruit and three servings of vegetables is classified as sufficient fruit and vegetable consumption.
422
S.-Y. Ho et al.
Food shops near home. The perceived presence of McDonald’s, KFC, Hong Kong-style fast-food shops, Chinese, Western and Hong Kong-style restaurants and 24-hour convenience stores near home was separately assessed. The students were asked whether each food shop was available within a fiveminute walking distance from home. A five-minute rather than a 15-minute walking distance (29) was used to define neighbourhood boundaries because Hong Kong is densely populated with high land use mix. The perceived availability of fast-food shops (McDonald’s, KFC, Hong Kong-style fast-food shops), restaurants (Chinese, Western and Hong Kong-style), and convenience stores near home was defined as reporting at least one specific food shop of that type near home. These three variables had acceptable test-retest agreements of 81.1%, 78.1% and 77.3%, respectively. Socio-demographic factors. Age, sex, highest parental education (primary, secondary or tertiary), and perceived family affluence (low, middle or high) were treated as potential moderators in the analyses. The perceived family affluence and highest parental education had good and excellent test-retest agreements of 79.4% and 95.7%, respectively. Both of these variables were correlated with housing size, possession of a car, and hiring a maid (r⫽0.156 to 0.348, mean⫽0.237, all p for trend⬍0.001). Weight status. For adjustment of weight status, ageand sex-standardized BMI z-scores were derived (30), based on self-reported weight and height. Alternative adjustment based on BMI-derived weight status (31) did not alter the results. Hence, weight status was treated as the continuous variable of BMI z-score. Data analyses Missing values for the highest parental education (14%) were imputed using multiple imputation methods taking the average of five imputed values. The estimates based on the imputed dataset approximated those based on the original dataset although their statistical significance was slightly strengthened due to the larger effective sample size. Only results based on the imputed dataset were reported. Logistic regression models with robust standard errors accounting for school clustering effects were used (Stata 9.0) to calculate adjusted odds ratios (OR) for predefined higher food intakes by the perceived availability of food shops. Potential multicollinearity between independent variables of perceived availability of food shops was examined by correlation
coefficients (ranging from r⫽0.17 to 0.39), i.e., none was greater than 0.8, which is suggestive of a problem (32). Separate models were used for each outcome of dietary intake, adjusting for age, sex, perceived family affluence, highest parental education, BMI z-score, and the perceived availability of different food shops mutually adjusting for each other. Interaction terms of food shops by socio-demographic factors (sex, age, perceived family affluence, and highest parental education) were added to the main effects to examine whether the relationship between food shops with eating behaviours were modified by socio-demographic factors. Bias may be introduced if, for each food intake, healthy and unhealthy eating adolescents differed in their ability to perceive the availability of food shops. This was examined by comparing the agreement between self-reported and objective measures of food shops in healthy and unhealthy eating groups. The presence of each type of food shops (fast-food shops/ restaurants/convenience stores) within a 400 m radius (five-minute walking distance (33)) from the home address of 62 randomly selected participants was measured objectively using Centamap, a commercial website with updated digital maps from the Lands Department, HKSAR (34). Kappa coefficient and kappa variance were used to calculate a Z-statistic for each dichotomised outcome of food intake (35). No significant difference in agreement between selfreported and objective measures in the two groups was observed for all outcomes (Z-statistics: 0.05 to 0.42⬍1.96; p⬎0.05). Results Most of the 24 796 students reported the presence of fast-food shops (65.5%), restaurants (78.0%) and convenience stores (59.6%) near their home, while only 18.7% and 31.8% consumed sufficient fruit and vegetables, respectively. Table I shows the basic characteristics of students, with 41.9% boys and a mean age of 14.5 years. Perceived family affluence was most commonly reported as middle (51.7%) and highest parental education as secondary (66.9%). Higher fruit, vegetables, and junk food/soft drinks consumption was associated with high perceived family affluence. Table II shows that adolescents reporting convenience stores near their home had 15% (95% confidence interval [CI]: 8%–23%) and 10% (4%–17%) greater odds of consuming more high-fat foods and junk food/soft drinks, respectively. Adolescents who reported fast-food shops near home were 10% (4%– 16%) and 10% (3%–17%) more likely to consume more high-fat foods and junk food/soft drinks, respectively. The perceived availability of restaurants was
2 293 (13.7) 11 118 (66.2) 3 378 (20.1)
922 (11.8)∗∗∗ 5 094 (65.2) 1 798 (23.0)
not significantly associated with consuming high-fat foods and junk food/soft drinks. Sensitivity analyses, using a lower cut-off value of ⱕthree days/month for high-fat foods or junk food/soft drinks gave similar results (data not shown). Table III shows that adolescents with restaurants near home were 17% (9%–24%) less likely to eat sufficient fruit and 13% (6%–20%) less likely to eat sufficient vegetables. Only 12% of students ate sufficient amounts of both and those who reported restaurants near home were 23% (16%–29%) less likely to be so. No significant association between the perceived availability of convenience stores and fastfood shops with fruit and vegetable intakes was observed. Table IV shows the significant interaction effects between socio-demographic factors and perceived availability of food shops on dietary intake. Lower fruit consumption and combined consumption with vegetables were associated with the perceived availability of restaurants only in adolescents with low and middle levels of family affluence, but not in those with high family affluence (p for interaction⫽0.02). Similarly, junk food/soft drinks intake was significantly higher in the presence of fast-food shops only among adolescents with low family affluence (p for interaction⫽0.048), but not higher family affluence. Moreover, the positive association between fast-food shops and consuming more junk food/soft drinks was found in boys, but not in girls (p for interaction⫽0.024). No other significant interaction was observed between socio-demographic factors and the perceived availability of food shops.
1 134 (13.6) 5 432 (65.2) 1 766 (21.2)
2 077 (12.8) 10 716 (66.2) 3 395 (21.0)
2 662 (13.3) 13 290 (66.5) 4 038 (20.2)
544 (11.9)∗∗∗ 2 904 (63.3) 1 138 (24.8)
276 (35.3)∗∗∗ 4 011 (51.3) 1 043 (13.4) 6 349 (37.8) 8 716 (51.9) 1 724 (10.3) 7 552 (37.8) 1 0382 (51.9) 2 056 (10.3) 5 844 (36.1)∗∗∗ 8 485 (52.4) 1 859 (11.5) 3 210 (38.5) 4 216 (50.6) 906 (10.9)
1 538 (33.5)∗∗∗ 2 335 (50.9) 713 (15.6)
⫺0.09 (0.13)∗∗∗ ⫺0.17 (0.11) ⫺0.20 (0.10) ⫺0.18 (0.11)∗∗∗ ⫺0.09 (0.13)
0.06 (0.15)∗∗∗
3 442 (44.1)∗∗∗ 6 844 (40.8) 8 217 (41.1) 6 736 (41.6)
2 136 (13.1) 1 0705 (65.7) 3 445 (21.2) significant difference between groups.
Relationship between perceived availability of food shops with consumption of high-fat foods and junk food/soft drinks
∗∗∗p⬍0.001;
5 977 (36.7) 8 451 (51.9) 1 858 (11.4)
⫺0.16 (0.11)
6 789 (41.7)
14.6 (1.86) 14.6 (1.86)∗∗∗ 14.4 (1.87)
423
Discussion
Age, mean (standard deviation, SD) (year) 14.5 (0.11) 14.4 (1.86) Sex, n (% boys) 10 379 (41.9) 3 464 (42.2) BMI z-score, mean (SD) ⫺0.15 (0.10) ⫺0.13 (0.13) Perceived family affluence, n (column %) Low 9 175 (37.0) 3 084 (37.5) Middle 12 825 (51.7) 4 235 (51.5) High 2 796 (11.3) 900 (11.0) Highest parental education, n (column %) Primary 3 241 (13.0) 1 066 (13.0) Secondary 16 330 (66.9) 5 445 (66.2) Tertiary 5 225 (21.1) 1 708 (20.8)
14.6 (1.86)∗∗∗
3 508 (42.1)
2 045 (44.6)∗∗∗
14.5 (1.87) 14.5 (1.86) 14.3 (1.85)∗∗∗
Sufficient (N⫽7 814) Insufficient (N⫽16 789) Insufficient (N⫽19 990)
Junk food/soft drinks
Moderate/high (N⫽16 188) Moderate/high (N⫽16 286)
High-fat foods
Low (N⫽8 219) Total (N⫽24 796)
Table I. Basic characteristics of participants by dietary habits.
Low (N⫽8 332)
Fruit
Sufficient (N⫽4 586)
Vegetables
Environment and dietary intakes
Previous studies found that a higher intake of fat and soft drinks was associated with eating in fast-food shops (15–17). Similarly, we found a positive association between the perceived availability of neighbourhood fast-food shops and convenience stores with adolescent intakes of high-fat foods and junk food/soft drinks. Although the variance explained (shown by the Nagelkerke R-square) by the models was small, our findings suggest the potential influence of the environment on adolescent dietary intake. Future studies should examine the roles of personal (e.g., food preference), behavioural (e.g., diningout), and environmental (e.g., availability of food at home) factors as mediators of the relationship between the availability of food within the built environment and adolescent dietary intakes.
424
S.-Y. Ho et al.
Table II. Prevalence of food shops and adjusted odds ratios for moderate/high consumption of high-fat foods and junk food/soft drinks. High-fat foods
Convenience stores (yes) Fast-food shops (yes) Restaurants (yes) Nagelkerke R-square
Junk food/soft drinks
Low#(N⫽8 219) n (%)
Moderate/ high#(N⫽16 286)n (%)
Adjusted OR (95% CI)
Low#(N⫽8 332) n (%)
Moderate/ high#(N⫽16 188)n (%)
Adjusted OR (95% CI)
4 693 (57.1)
9 956 (61.1)
1.15 (1.08–1.23)∗∗∗
4 800 (57.6)
9 861 (60.9)
1.10 (1.04–1.17)∗∗
5 202 (63.3)
10 851 (66.6)
1.10 (1.04–1.16)∗∗
5 285 (63.4)
10 764 (66.5)
1.10 (1.03–1.17)∗∗
6 349 (77.2)
12 787 (78.5) 8.0%
0.98 (0.92–1.04)
6 437 (77.3)
12 725 (78.6) 6.1%
1.00 (0.93–1.08)
OR: Odds ratios; CI: Confidence interval. 1Adjusted for other food shops in the Table, age, sex, perceived family affluence, highest parental education, and body mass index z-score. ∗p⬍0.05, ∗∗p⬍0.01, ∗∗∗p⬍0.001. Reference group: no respective food shop. #Low:ⱕone day/week; moderate/high consumption: ⱖtwo days/week.
Relationship between perceived availability of food shops with fruit and vegetable intake Restaurants but not fast-food shops or convenience stores were negatively associated with fruit and vegetable intake. This seems to contradict previous Western reports (11,17). This inconsistency; however, could be due to differences in eating habits. In Western countries, adolescents typically consume fast food frequently (36–38). The patronage of nearby fast-food shops, which offer mainly high-fat and energy-dense foods, may therefore result in lower intake of fruit and vegetables. In Hong Kong, people who dine out generally prefer restaurants to fast-food shops (39), and this may explain why the detrimental effect on fruit and vegetable consumption was observed for the presence of neighbourhood restaurants but not fast-food shops and convenience stores. It has been argued that the availability of fastfood shops, restaurants, and convenience stores, one aspect of the “obesogenic” environment promotes unhealthy food intake but discourages healthy food consumption (22,25). Our findings support this concept. Future studies should examine how the built environment influences adolescents’ dietary intake through personal and behavioural factors and the availability of foods within the home.
Perceived family affluence and sex as moderators Family affluence significantly moderated the association between the perceived availability of restaurants and fruit intake, such that the negative association was observed only in low or middle but not the high affluence group. A probable explanation is that less affluent adolescents would be less likely to consume a decent meal in restaurants that comes with complimentary fruit. Regardless of the amount of fruit eaten in restaurants, affluent adolescents are also
more likely to have fruit available at home (40). On the contrary, as fruit is not cheap and generally regarded as inessential, low income parents would be less likely to buy and encourage their children to eat more fruit after having dinner in restaurants. A significant association was observed between the perceived availability of neighbourhood fast-food shops and junk food/soft drinks consumption only in low family affluence group. This group may not be able to afford eating in venues with a healthier menu and would thus be more likely to frequent fast-food shops if they were available. In Hong Kong preparing dinner at home may not be cheaper than buying set meals, which are typically bundled with soft drinks at fast-food shops. A further limitation to dining at home occurs in families where both parents work long hours, which prohibits cooking at home. In Hong Kong, maids are common place in middleclass families and are responsible for food preparation. However, maids are less affordable to poorer families and thus increase the requirement for outside dining. We found that only boys but not girls who reported fast-food shops near their home were likely to consume more junk food/soft drinks. Other studies have reported that body dissatisfaction and dieting were more prevalent in girls (41–42), whereas visits to fast-food shops and consumption of soft drinks were more prevalent in boys (16,18). This suggests that although the perceived availability of fast-food shops was similar in boys (36%) and girls (31%), girls may consume less soft drinks than boys due to greater concern over their weight status. Another possible explanation is that boys and girls might have consumed junk food/soft drinks from different sources not measured in the present study (18). Future studies should further examine how the environment influences the dietary intakes differently in boys and girls.
7.0% 12.4%
OR: Odds ratios; CI: Confidence interval. 1Adjusted for other food shops variables in the Table, age, sex, perceived family affluence, highest parental education, and body mass index z-score. ∗p⬍0.05, ∗∗p⬍0.01, ∗∗∗p⬍0.001. Reference group: no respective food shop. #Sufficient: ⱖtwo servings; insufficient: ⬍two servings. ##Sufficient: ⱖthree servings; insufficient:
