Association between neighbourhood fast-food and full-service restaurant density and body mass index: a cross-sectional study of Canadian adults.

Hollands, Simon ; Campbell, M. Karen ; Gilliland, Jason 等

La traduction du resume se trouve a la fin de l'article.

Adult body mass index (BMI) in Canada and in other developed nations has increased dramatically over the last three decades, and obesity is now considered an epidemic. (1) BMI is defined as weight (kg) divided by [height.sup.2] ([m.sup.2]), and obesity is defined as a BMI of 30 or higher. It is argued by some that obesity can be attributed to modern built environment features that promote unhealthy eating and sedentary lifestyle. (2) One aspect of the built environment is the exposure to fast-food restaurants seen in many neighbourhoods. Fast-food meals are characterized by large portion sizes, high levels of saturated fat and high caloric density. (3) Thus, frequent fast-food consumption is a risk factor for obesity. (4)

Exposure to a fast-food environment is typically measured by the density of fast-food outlets in a defined local neighbourhood or the distance to the nearest fast-food outlet. It is hypothesized that higher densities of neighbourhood-level fast-food outlets contribute to an increased prevalence of obesity. One mechanism by which this hypothesis is argued is the existence of food deserts, whereby residents of socio-economically disadvantaged neighbourhoods tend to have poor access to healthy food vendors (i.e., supermarkets, grocery stores, fruit and vegetable markets). (5) In the United States and internationally, a number of studies have found a positive association between the proximity or density of fast-food restaurants and differences in BMI or risk of obesity; (6-9) some studies have found a negative association, (10) and others have found no association. (11-14) In the Canadian context, there is very little evidence on this topic. Two ecological studies have shown an association between area-level fast-food density and obesity in Canada. (15,16) Only two Canadian studies have analyzed individual-level data, but each focused only on children in a single urban centre (Edmonton, AB; London, ON). (17,18) Both studies found modest associations between fast-food density and childhood obesity. This study aims to address the gap in the Canadian literature and contribute to the evolving international literature on this topic. Specifically, we examine the influence of the local food environment (neighbourhood-level fast-food and full-service restaurant densities) on BMI for the first time, using individual-level nationally representative survey data from Canada.

METHODS

Data and variables

The data for this study came from the Canadian Community Health Survey (CCHS) conducted by Statistics Canada. The CCHS is designed to collect information related to health status, health care utilization and health determinants from a nationally representative sample of the Canadian general population. Details on survey methodology can be found elsewhere. (19) The CCHS 2007-08 confidential master file (n=131,959) was used to obtain individual BMI (weight [kg]/[height.sup.2] [[m.sup.2]]), lifestyle and socio-demographic factors. The CCHS master file was accessed through Statistics Canada's Research and Data Centre, which allowed access to respondents' postal codes and unsuppressed data for variables such as income and age.

Adults (aged 18 to 65) who resided in one of Canada's 10 provinces were included in our analyses (n=89,733). Residents of the three territories of Canada were excluded because of their distinct demographic and geographic features. Respondents who were pregnant or did not report height or weight (n=4,321), had extreme BMI values (<10 or >70) (n=26), or were still breastfeeding (n=1,045) were excluded, which left 84,341 eligible respondents for analysis. For self-reported surveys, it has been suggested that individuals systematically under-report weight and over-report height, which can lead to biased BMI estimates. (20) Therefore, a validated error correction factor, (20) developed using previous CCHS survey data, was applied as follows: male [BMI.sub.corrected] = -1.08 + 1.08 ([BMI.sub.self-report]); female [BMI.sub.corrected] = -0.12 + 1.05 ([BMI.sub.self-report]).

Various lifestyle characteristics (smoking, alcohol use, physical activity, fruit and vegetable consumption, sedentary activity), socio-economic status (immigration, race, labour market, income, education, food security) and demographic characteristics (sex, marital status, having children, urban region and province) were identified as confounding variables. These variables were available in the CCHS and considered as confounders because they were associated with either access to or consumption of fast food and BMI in the literature. These variables, shown in Table 1, were controlled for in the regression analyses so that the independent effects of the primary exposures could be teased out. Neighbourhood-level factors were also shown to confound the relationship in question. (21) Therefore, neighbourhood-level socio-demographic characteristics (transportation to work, marriage, income and education) at the level of the census dissemination area (DA) were controlled. DAs are the smallest geographical unit, consisting of 400 to 700 people, for which data are available in Canada. (22)

The names and geographic locations of all restaurants (standard industry classification code: 5812-08) were purchased from the infoCanada[R] business database. This information was geocoded and linked with the CCHS data. The details of the process have been published elsewhere. (15) Fast-food (or limited service) chain restaurants were defined as those food establishments (including general coffee outlets) that provide services to customers on the basis of food being ordered and paid for before it is eaten or taken out. Full-service restaurants were defined as those establishments that provide food services on the basis of patrons being served food and paying after their meal (i.e., servers are available). This classification system had been used previously. (15)

The names provided in the infoCanada business database were cross-referenced with business names published in the 2008 Canadian Restaurant Directory. (23) There were 19,524 fast-food restaurants in our database vs. 21,418 reported in the 2008 Directory and 3,894 full-service restaurants in our database vs. 3,823 in the Directory. We identified 4,534 records classified as independent pizza outlets, which were included in our fast-food restaurant category. All other non-chain restaurants were classified as other (n=41,972). These non-chain restaurants were independently owned and very specific to the local community. For example, many ethnic restaurants belonged to this category. It was impossible to classify these restaurants into fast-food or full-service with any verifiable source or directory. For each restaurant category (fast-food, full-service and non-chain restaurants), the number of outlets per forward sortation area (FSA) was divided by the corresponding FSA population (per 10,000) based on the 2006 Census to construct our restaurant density variables. FSAs are geographical areas that consist of the first three digits of Canadian postal codes.

Statistical analysis

The association between BMI and the density of fast-food and full-service restaurants was analyzed using ordinary least squares regression (OLS) with and without control for confounding variables. Sampling weights were applied to all descriptive and regression analyses. Robust standard errors were used to account for unknown forms of heteroskedasticity in the data, which were clustered at the FSA level. Stratified analyses were performed by sex and two levels of geographic classification: census metropolitan areas (CMA) and non-CMA. A CMA consists of neighbouring municipalities and has a total population of at least 100.000 (in which 50,000 or more live in the urban core). (22) Stratified analysis by CMA and non-CMA is relevant for two reasons. First, the eating behaviour of individuals living in urban jurisdictions is expected to be different from that of rural or semi-urban residents because of differential time constraints they face. Second, our restaurant measures are more meaningful in urban areas than rural areas. Exclusion of missing values of all variables resulted in a final sample size of 72,660. The software STATA 12[c] was used for all analyses.

RESULTS

A total of 1,558 relevant FSAs across Canada had on average 7 fast-food, 1 full-service and 13 other non-chain restaurants per 10,000 individuals. The mean (standard deviation [SD]) age of respondents was 42.2 (13.2) years. The mean (SD) BMI of the sample was 26.97 (5.34), males and residents in non-CMAs having statistically higher BMI (p<0.001 for both) than their female and CMA counterparts. The weighted proportion of males and females was about equal. Roughly 80% identified themselves as White and Canadian citizens, and over three quarters were employed. Forty-seven percent of the respondents reported being physically inactive, and the majority were married and had postsecondary education (Table 1).

In both model 1 (bivariate) and model 2 (adjusted) OLS regression analyses, fast-food density was statistically significant in the positive direction, and full-service and other non-chain restaurant density was statistically significant in the negative direction with respect to BMI (Table 2). After the confounding effects of lifestyle, socio-demographic and neighbourhood factors had been controlled for, the estimated coefficients of restaurant density variables were lower (Table 2). The adjusted estimated regression coefficient for fast-food density was 0.031 (95% confidence interval [CI]: 0.017 to 0.045). This finding suggests that for an average individual 1.76 metres tall (5' 10") weighing 72.5 kg (160 lb), an increase of 10 fast-food restaurants (per 10,000 population) is associated with a positive weight difference of 1 kg (weight difference [kg]=[beta]*density*[m.sup.2]). The negative association found between full-service restaurants and BMI ([beta]= -0.06 [95% CI: -0.11 to -0.013]) shows that an increase of 10 full-service restaurants per 10,000 population is associated with a negative weight difference of 1.9 kg for the same average person.

In model 3 (stratified by sex), the estimated coefficients for fast-food and other restaurant density for males and females were similar to the overall results. Full-service restaurant density was not significant for women or men (Table 3). Model 4 (CMA/non-CMA) (Table 3) shows that the magnitude of the association between fast-food availability and BMI exists only in urban areas. The interaction between sex and urbanicity is presented in model 5, in which we found the strongest association between the food service environment and BMI (Table 4). For males, the estimated coefficient of fast-food density did not differ substantially when further stratified by CMA ([beta]=0.032; 95% CI: 0.001 to 0.063), but the estimated coefficients of full-service restaurant density were amplified ([beta]= -0.1 (-0.19 to-0.0018). For men, an increase of 10 full-service restaurants (per 10,000 population) was associated with a decrease in BMI by one point. The association between fast-food density and BMI was stronger in women residing in a CMA compared with the overall and sex-only stratification. The estimated regression coefficient of 0.041 (95% CI: 0.01 to 0.072) can be interpreted as a positive weight difference of 1.14 kg for a woman 58 kg (128 lb) in weight and 1.67 m (5' 6") in height with an additional 10 fast-food outlets (per 10,000 population) in an FSA. Full-service density was not statistically significant for women in the interaction model.

As far as the direction of the association for the other covariates is concerned, the results were in the expected direction. For example, on average BMI increased with age (p<0.001) and was higher for those who were married (p<0.01) and those who had no education (p<0.001). Recent immigrants and visible minorities had lower BMI, but mean BMI increased with duration of time since immigration. Several confounders were more important in stratified analyses. For example, income adequacy was not statistically significant in the full sample, as shown in the Appendix. However, when stratified by sex, income became significant and in the opposite directions for males and females (high vs. low income: [beta]=0.56, p<0.001, for males; [beta]= -0.82, p<0.001, for females). This suggests that for males, a high income is associated with a 0.56 points higher BMI on average compared with low-income male counterparts, whereas for women, a high income is associated with an average BMI that is 0.82 points lower than that of low-income females. Similarly, men who were married had a higher BMI than those who were single, but the effect of marriage was not significant in females (married vs. single: [beta]=0.63, p<0.01, for males; [beta]= -0.2, p>0.05, for females).

DISCUSSION

This study found that adult BMI was higher in areas in Canada that had a greater density of fast-food restaurants. An inverse association was seen with full-service restaurant density. The findings were robust even after the influence of individual-level lifestyle, socio-economic and demographic factors, as well as neighbourhood-level socio-economic factors, had been controlled for. The observed associations were predominantly found in Canada's CMAs, where the magnitude of associations for men and women differed. For a female of average height (1.67 m), a mean difference of 1.14 kg was estimated between FSAs that differed by 10 fast-food outlets (per 10,000 population). For men living in CMAs, the negative association between full-service restaurant density and BMI was more pronounced. On average, an FSA with 10 additional full-service restaurants corresponded to a negative weight difference of 3 kg (for males 1.76 metres tall).

This study adds to a growing body of literature focusing on the contextual effects of the built environment that promote unhealthy eating and sedentary lifestyle. (2) Exposure to fast-food outlets is argued by some to be a primary contributor in this regard. (15,24) Ecological studies, in which fast-food restaurant density has been linked to higher obesity rates, have offered evidence supporting the findings shown in this study. (25) In Canada, a recent study demonstrated an association between BMI and fast-food restaurant density using FSA-level data, citing the need for further research using individual-level data. (15) When individual-level data have been considered, the results so far have been largely mixed. Many studies have found a statistically significant positive association between fast-food restaurant density and adult obesity, (6-9) though others have not, (11-14) and some have actually reported an inverse relationship. (10) Where a significant association was not observed, often the data were beset by a small sample or a potential lack of neighbourhood restaurant data variability. For instance, Simmons et al. (14) studied only 7 towns, and Wang et al. (13) used only 82 neighbourhoods. In this study, over 1,500 FSAs across a large sample of nationally representative adults were used, and the results were not dramatically different after accounting for confounding effects.

This study offered a comprehensive analysis of the local food-service environment with the inclusion of full-service restaurants and other non-chain restaurants in the analyses. As with Mehta and Chang in the US, (7) we found that the density of full-service restaurants had an inverse association with BMI in Canada. The opposite effects seen between fast-food and full-service restaurants may highlight the importance of consumer demand for convenience in developed nations. Dining at full-service restaurants is often for social or entertainment purposes, and purchasing decisions are not as likely to be made spontaneously, driven by time constraints and convenience. Full-service restaurants also have more restrictive hours than chain fast-food restaurants; in fact, irregular eating patterns, especially at night, are associated with increased obesity risk. (26) Overall, greater full-service restaurant density may also represent a more advantageous eating environment. One study found that consumers who valued healthy foods were 29% more likely to choose full-service establishments over fast-food outlets. (27) A negative effect was also seen for non-chain restaurant density, though the strength of the association was less than that for full-service restaurants. The findings suggest that, as a risk factor for obesity, on average non-chain restaurants may have some characteristics similar to those of full-service restaurants rather than of chain fast-food outlets.

Limitations

The cross-sectional study design limits the ability to draw causal inferences from observed associations. A study in which both BMI and area-level restaurant density were tracked longitudinally could be more useful, as the effect of changes in restaurant density on changes in BMI or a measure of obesity could be studied in order to ascertain causal associations. Another potential drawback arises from BMI being derived from self-reported data. Although an error correction factor was applied, point estimates may still be biased in the downward direction as a result of under-reporting of weight. We think this bias might have underestimated the strengths of the associations between restaurant density and BMI in this study. Another limitation of CCHS data is that First Nations people living in a First Nations community are excluded from the sampling frame. With regard to the classification of restaurants, no rigid definition for what constitutes a fast-food restaurant exists in the literature. The drawback of our approach is a certain degree of inevitable misclassification, as a number of independent establishments may offer fast-food services. This may be especially true for independent ethnic restaurants.

Another issue is the decision to use FSA as the relevant neighbourhood-level geographic unit for exposure to the local food-service environment. Other studies have used more exact measures, such as buffer zones of varying areas around individual places of residence. (13,24) In our case, the use of buffer zones was not possible because the smallest unit of geographic identification available in the CCHS data was postal codes. However, constructing restaurant density variables at a lower level of geographic classification, such as six-digit postal codes or dissemination areas, may not be meaningful since individuals are more likely to consume restaurant foods in the vicinity of their residence and workplace.

Despite the above limitations, this study has a number of strengths. Most important, it was carried out using data from a large nationally representative sample of Canadian adults. Moreover, we had access to reliable geographic data on the locations of all restaurants in Canada, from which we were able to construct area-level restaurant density variables. This study was able to control for a rich set of socio-economic and neighbourhood-level confounding variables.

CONCLUSIONS

This research is the first to investigate the association of fast-food and full-service restaurant density with BMI using individual-level data from a nationally representative survey from Canada. These findings are important, as fast-food availability is potentially a practical policy lever. For instance, zoning bylaws could be implemented at the municipal level to regulate the number and density of fast-food restaurants, and their proximity to schools and hospitals, or to institute an outright ban in certain areas. (28,29) Several US cities have begun adopting similar bylaws recently. (28) Epidemiologic evidence is needed to consider and understand the effectiveness of these types of initiatives in Canada.

BMI was the primary outcome examined in this study, but the adverse health effects of fast food on health outcomes may be wide-ranging. Among other health problems, exposure to fast-food restaurants has been associated with mortality and hospital admissions for acute coronary events in Ontario, (30) further highlighting the overarching health risks associated with exposure to fast food.

Appendix. Efficients of All Covariates (Model 2)

Variable                             Model 2 (Multivariable)
                                            p: (95% CI)

Fast-food
  restaurant density                0.031 *** (0.017 to 0.045)
Full-service
  restaurant density               -0.061 ** (-0.110 to -0.013)
Other restaurant density           -0.014 *** (-0.019 to -0.008)
Age                                 0.263 *** (0.230 to 0.296)
Geography
  Urban (>500K pop)                -0.323 *** (-0.500 to -0.147)
  Non-urban (ref)                               --
Sex
  Female                           -1.524 *** (-1.649 to -1.399)
  Male (ref)                                    --
Marital status
  Married                            0.221 ** (0.051 to 0.392)
  Widowed/separated/divorced          0.035 (-0.214 to 0.283)
  Single (ref)                                  --
Education
  Secondary                        -0.309 ** (-0.544 to -0.074)
  Some post-secondary              -0.363 *** (-0.634 to -0.092)
  Post-secondary                   -0.552 *** (-0.769 to -0.334)
  No education (ref)                            --
Immigration
  Immigrant (<10 y in Canada)      -1.459 *** (-1.778 to -1.140)
  Immigrant (>10 y in Canada)      -0.486 *** (-0.701 to -0.270)
  Canadian (ref)                                --
Minority
  Non-White                        -1.004 *** (-1.278 to -0.729)
  Race missing                      0.579 *** (0.270 to 0.888)
  White (ref)                                   --
Labour market
  Student                          -0.482 *** (-0.776 to -0.189)
  No work                             0.043 (-0.133 to 0.219)
  Unable to work                      0.155 (-0.296 to 0.606)
  Working (ref)                                 --
Income adequacy
  Mid low                             0.015 (-0.177 to 0.207)
  Mid                                0.163 * (-0.020 to 0.345)
  Mid high                            0.115 (-0.075 to 0.305)
  High                               -0.046 (-0.235 to 0.143)
  Income missing                      0.015 (-0.177 to 0.207)
  Low (ref)                                     --
Child age (0-6)
  Yes                               0.261 *** (0.068 to 0.455)
  No (ref)                                      --
Child age (6-11)
  Yes                                -0.093 (-0.272 to 0.087)
  No (ref)                                      --
Food security
  Moderate                          0.840 *** (0.528 to 1.153)
  Insecure                            0.422 (-0.088 to 0.933)
  Secure (ref)                                  --

Variable                             Model 2 (Multivariable)
                                         [beta]: (95% CI)
Fruit and vegetable consumption
  Medium                             -0.064 (-0.187 to 0.059)
  High                             -0.444 *** (-0.707 to -0.181)
  Low (ref)
Smoking status                                  --
  Daily                            -1.080 *** (-1.259 to -0.901)
  Occasional                         -0.069 (-0.345 to 0.208)
  Former daily                      0.443 *** (0.282 to 0.605)
  Former occasional                   0.043 (-0.123 to 0.210)
  Never (refs)
Drinking habits                                 --
  Occasional                        0.533 *** (0.294 to 0.773)
  Regular                          -0.688 *** (-0.880 to -0.496)
  Never (refs)
Physical activity                               --
  Moderate                          0.471 *** (0.324 to 0.618)
  Inactive                          1.099 *** (0.957 to 1.240)
  Active (ref)
Sedentary activity                              --
  Moderate                          0.232 *** (0.077 to 0.386)
  High                              0.852 *** (0.691 to 1.013)
  Very high                         1.316 *** (1.123 to 1.510)
  Low (refs)
Province                                        --
  NL                                 0.502 ** (0.117 to 0.887)
  PE                                 0.299 ** (0.048 to 0.550)
  NS                                 0.291 ** (0.031 to 0.551)
  NB                                  0.196 (-0.068 to 0.459)
  QC                               -0.490 *** (-0.696 to -0.284)
  MB                                  0.219 (-0.092 to 0.530)
  SK                                0.502 *** (0.231 to 0.773)
  AB                                 0.228 ** (0.005 to 0.451)
  BC                               -0.682 *** (-0.876 to -0.487)
  ON (ref)                                      --

Neighbourhood-level variables
Proportion in Dissemination Area

Drive to work                        0.495 * (-0.021 to 1.010)
Married                               0.094 (-0.585 to 0.772)
Low-income families                   0.557 (-0.147 to 1.260)
Visible minority                   -0.731 *** (-1.231 to -0.230)
Low education                       2.114 *** (1.486 to 2.742)

* p < 0.1.

** p < 0.05.

*** p < 0.01.

Acknowledgements: This paper uses confidential microdata files from Statistics Canada's Canadian Community Health Survey, 2007-08, and from the restaurant database from info Canada. All data analyses were conducted at the University of Western Ontario Research Data Centre. Funding for this research by the Canadian Institutes of Health Research operating grant (reference number: MOP-97763) is gratefully acknowledged. This is a substantially revised version of Simon Hollands' thesis chapter submitted to the University of Western Ontario. The views expressed, however, are those of the authors and do not necessarily reflect the views of any affiliated organization.

REFERENCES

(1.) Hedley AA, Ogden CL, Johnson CL, Carroll MD, Curtin LR, Flegal KM. Prevalence of overweight and obesity among US children, adolescents, and adults, 1999-2002. JAMA 2004;291(23):2847-50.

(2.) Feng J, Glass TA, Curriero FC, Stewart WF, Schwartz BS. The built environment and obesity: A systematic review of the epidemiologic evidence. Health Place 2010;16(2):175-90.

(3.) Prentice AM, Jebb SA. Fast foods, energy density and obesity: A possible mechanistic link. Obes Rev 2003;4(4):187-94.

(4.) Pereira MA, Kartashov AI, Ebbeling CB, Van Horn L, Slattery ML, Jacobs DR Jr, et al. Fast-food habits, weight gain, and insulin resistance (the CARDIA study): 15-year prospective analysis. Lancet 2005;365(9453):36-42.

(5.) Larsen K, Gilliland J. Mapping the evolution of 'food deserts' in a Canadian city: Supermarket accessibility in London, Ontario, 1961-2005. Int J Health Geogr 2008;7:16.

(6.) Morland KB, Evenson KR. Obesity prevalence and the local food environment. Health Place 2009;15(2):491-95.

(7.) Mehta NK, Chang VW. Weight status and restaurant availability: A multilevel analysis. Am J Prev Med 2008;34(2):127-33.

(8.) Li F, Harmer P, Cardinal BJ, Bosworth M, Johnson-Shelton D. Obesity and the built environment: Does the density of neighborhood fast-food outlets matter? Am J Health Promot 2009;23(3):203-9.

(9.) Dunn R. Obesity and the availability of fast-food: An instrumental variables approach (March 31 2008). International Health Economics Association 6th World Congress, 2007: Explorations in Health Economics Working Paper. Available at: http://ssrn.com/abstract=989363.

(10.) Pearce J, Hiscock R, Blakely T, Witten K. A national study of the association between neighbourhood access to fast-food outlets and the diet and weight of local residents. Health Place 2009;15(1):193-97.

(11.) Rundle A, Neckerman KM, Freeman L, Lovasi GS, Purciel M, Quinn J, et al. Neighborhood food environment and walkability predict obesity in New York City. Environ Health Perspect 2009;117(3):442-47.

(12.) Casey AA, Elliott M, Glanz K, Haire-Joshu D, Lovegreen SL, Saelens BE, et al. Impact of the food environment and physical activity environment on behaviors and weight status in rural U.S. communities. Prev Med 2008;47(6):600-4.

(13.) Wang MC, Kim S, Gonzalez AA, MacLeod KE, Winkleby MA. Socioeconomic and food-related physical characteristics of the neighbourhood environment are associated with body mass index. J Epidemiol Community Health 2007;61(6):491-98.

(14.) Simmons D, McKenzie A, Eaton S, Cox N, Khan MA, Shaw J, et al. Choice and availability of takeaway and restaurant food is not related to the prevalence of adult obesity in rural communities in Australia. Int J Obes (Lond) 2005;29(6):703-10.

(15.) Hollands S, Campbell MK, Gilliland J, Sarma S. A spatial analysis of the association between restaurant density and body mass index in Canadian adults. Prev Med 2013;57(4):258-64.

(16.) Cash S, Goddard E, Lacanilao R. Fat taxes and health outcomes: An investigation of economic factors influencing obesity in Canada. Project report. Edmonton, AB: Faculty of Agriculture & Forestry, and Home Economics, University of Alberta, 2007.

(17.) Spence JC, Cutumisu N, Edwards J, Evans J. Influence of neighbourhood design and access to facilities on overweight among preschool children. Int J Pediatr Obes 2008;3(2):109-16.

(18.) Gilliland JA, Rangel CY, Healy MA, Tucker P, Loebach JE, Hess PM, et al. Linking childhood obesity to the built environment: A multi-level analysis of home and school neighbourhood factors associated with body mass index. Can J Public Health 2012;103(Suppl 3)(9):S15-21.

(19.) Statistics Canada. Canadian Community Health Survey--Annual Component (CCHS). Detailed information for 2007 (cycle 4.1). 2008. Available from: http://www.statcan.gc.ca/cgi-bin/imdb/p2SV.pl?Function=getSurvey&SurvId=3226&SurvVer=1&InstaId= 15282&InstaVer=4&SDDS=3226&lang=en&db=imdb&adm=8&dis=2 (Accessed March 29, 2012).

(20.) Gorber CS, Shields M, Tremblay MS, McDowell I. The feasibility of establishing correction factors to adjust self-reported estimates of obesity. Health Rep 2008;19(3):71-82.

(21.) Morland K, Wing S, Diez Roux A, Poole C. Neighborhood characteristics associated with the location of food stores and food service places. Am J Prev Med 2002;22(1):23-29.

(22.) Statistics Canada. Census Dictionary. Catalogue no 98-301-X2011001. 2011:114-17.

(23.) Monday Report on Retailers. Directory of Restaurant and Fast Food Chains in Canada. Toronto, ON: Monday Report on Retailers, 2010.

(24.) Spence JC, Cutumisu N, Edwards J, Raine KD, Smoyer-Tomic K. Relation between local food environments and obesity among adults. BMC Public Health 2009;9:192.

(25.) Chou SY, Grossman M, Saffer H. An economic analysis of adult obesity: Results from the Behavioral Risk Factor Surveillance System. J Health Econ 2004;23(3):565-87.

(26.) Stunkard AJ, Allison KC. Two forms of disordered eating in obesity: Binge eating and night eating. Int J Obes Relat Metab Disord 2003;27(1):1-12.

(27.) Stewart H, Blisard N, Jolliffe D. Let's eat out: Americans weigh taste nutrition and convenience. Econ Info Bull 2006 October;(EIB-19):16.

(28.) Eisenberg MJ, Atallah R, Grandi SM, Windle SB, Berry EM. Legislative approaches to tackling the obesity epidemic. CMAJ 2011;183(13):1496-500.

(29.) Mair JS, Pierce MW, Teret SP. The City Planner's Guide to the Obesity Epidemic: Zoning and Fast Food. The Centers for Law and the Public's Health: a Collaborative at Johns Hopkins and Georgetown Universities, 2005.

(30.) Alter DA, Eny K. The relationship between the supply of fast-food chains and cardiovascular outcomes. Can J Public Health 2005;96(3):173-77.

Simon Hollands, MSc, [1] M. Karen Campbell, PhD, [1-4] Jason Gilliland, PhD, [3-6] Sisira Sarma, PhD [1]

Author Affiliations

[1.] Department of Epidemiology & Biostatistics, Schulich School of Medicine & Dentistry, University of Western Ontario, London, ON

[2.] Department of Obstetrics and Gynecology, Schulich School of Medicine & Dentistry, University of Western Ontario, London, ON

[3.] Department of Paediatrics, Schulich School of Medicine & Dentistry, University of Western Ontario, London, ON

[4.] Children's Health Research Institute; Lawson Health Research Institute, London, ON

[5.] Department of Geography, University of Western Ontario, London, ON

[6.] School of Health Studies, University of Western Ontario, London, ON

Correspondence: Sisira Sarma, PhD, Kresge Bldg, Room K201, Department of Epidemiology & Biostatistics, Schulich School of Medicine & Dentistry, University of Western Ontario, London, ON N6A 5C1, Tel: 519-661-2111, ext. 87583, E-mail: ssarma2@uwo.ca

Conflict of Interest: None to declare.

Received: October 10, 2013

Accepted: March 1, 2014

Table 1. Descriptive statistics

Variable                                                 Value *
Individual level

BMI, mean (SD)
  Overall                                             26.97 (5.34)
  Males                                                27.59 (4.9)
  Females                                             26.31 (5.72)
  Residents of CMAs                                   26.61 (5.21)
  Residents of non-CMA                                27.79 (5.58)
Food service density, ([dagger]) mean (SD)
  Fast-food service (per 10,000 FSA population)        7.67 (7.47)
  Full-service restaurant service
    (per 10,000 FSA population)                        1.25 (1.76)
  Other restaurant service (per 10,000
    FSA population)                                   13.41 (15.42)
Smoking status, %
  Non-smoker (has never smoked)                            36%
  Daily smoker (smokes every day)                          21%
  Always occasional smoker
    (never a daily smoker or has
    smoked less than 100 cigarettes lifetime)              5%
  Current occasional smoker (former daily smoker)          23%
  Current non-smoker (former
    occasional or former daily)                            15%
Alcohol use, %
  Non-drinker (has not drunk in the past year)             15%
  Occasional drinker (drinks less
    than once per month)                                   15%
  Regular drinker (drinks once per month or more)          70%
Physical activity (leisure and transportation), %
  Active (daily energy expenditure >30 minutes)            27%
  Moderate (daily energy
    expenditure 20-30 minutes)                             26%
  Inactive (daily energy expenditure <15 minutes)          47%
Frequency of fruits & vegetables consumption, %
  Low (<5 times per day)                                   58%
  Middle (5-10 times per day)                              37%
  High (>10 times per day)                                 5%
Sedentary activity (excluding reading, e.g., TV,
computer games), %
  Low (0-9 hours per week)                                 20%
  Middle (10-20 hours per week)                            30%
  High (20-30 hours per week)                              34%
  Very high (30+ hours per week)                           16%
Immigration, %
  Canadian                                                 78%
  Immigrant (<10 years in Canada)                          7%
  Immigrant (>10 years in Canada)                          15%
Race, %
  White                                                    81%
  Other (any race reported other than "White")             16%
  Race not reported                                        3%
Labour market, %
  Work (full or part time)                                 78%
  Student (who does not work)                              4%
  Does not work                                            16%
  Permanently unable to work                               2%
Income decile, ([double dagger]) %
  Low 1st-2nd                                              15%
  Mid low 3rd-4th                                          16%
  Mid 5th-6th                                              18%
  Mid high 7th-8th                                         20%
  High 9th-10th                                            20%
  Income not reported                                      10%
Education, %
  Less than secondary                                      11%
  Secondary                                                17%
  Some post-secondary                                      10%
  Post-secondary                                           62%
Food security, %
  Secure (no, or one, indication of difficulty
    with income-related food access)                       93%
  Moderately secure (indication of compromise in
    quality and/or quantity of food consumed)              6%
  Insecure (indication of reduced food intake and
    disrupted eating patterns)                             2%
Sex, %
  Female                                                   49%
  Male                                                     51%
Marital status, %
  Currently married                                        64%
  Single, never married                                    26%
  Separated/divorced/widowed                               10%
Children, %
  Has a child aged <6                                      15%
  Has a child aged 6-11                                    16%
Region
  Urban region (population >500,000)                       49%

Neighbourhood Dissemination Area (DA)                  Mean of the
level                                                  proportion
                                                          (SD)

Drive to work (population in CCHS respondents'
  DA who drive or carpool to work)                     0.72 (0.17)
Married (population in CCHS respondents' DA
  over the age of 15 who are married)                  0.48 (0.14)
Low income (population in CCHS respondents' DA
  who meet Statistics Canada's low income cut-off)     0.12 (0.12)
Visible minority (population in CCHS respondents'
  DA who are visible minorities, i.e., non-White)      0.16 (0.22)
Low education (population in CCHS respondents'
  DA over the age of 15 who have less than a
  high-school education)                               0.23 (0.11)

* All values represent weighted percentages.

([dagger]) Density defined as per 10,000 forward sortation
area population, based on 2006 census population estimates.

([double dagger]) Income adequacy deciles are based on
provincial standardization and take into account both income
and household size.

CMA=census metropolitan area: total population of 100,000
(in which 50,000 or more live in the central core).

Table 2. Bivariate and multivariate association * between
BMI and restaurant density, by restaurant service type

                          Total sample: (95% CI)

Restaurant type        Model 1           Model 2
([dagger])           (Bivariate)      (Multivariable)
                                     ([double dagger])

Fast-food service    0.037 (0.022      0.031 (0.017
                      to 0.053)          to 0.045)
                     ([section])        ([section])

Full service        -0.116 (-0.17      -0.06 (-0.11
                      to -0.062)        to -0.013)
                     ([section])        ([section])

Other restaurant    -0.024 (-0.03     -0.014 (-0.019
  service             to -0.016)        to -0.008)
                     ([section])        ([section])

* Survey sampling weights provided by Statistics Canada were
used in all regressions.

([dagger]) Density: outlets per 10,000 forward sortation
area population based on the 2006 Canadian Census.

([double dagger]) Individual-level factors controlled for:
demographic characteristics, socio-economic status,
lifestyle and geographic characteristics, dissemination
area. The estimated coefficients for these covariates are
presented in the Appendix.

([section]) Statistically significant at p<0.05.

Table 3. Multivariate association* between BMI and
restaurant density stratified by sex and urbanicity, by
restaurant service type

Restaurant             Model 3 by sex: [beta] (95% CI))

type                     Males                Females
([dagger])

Fast-food                0.029                 0.03
service           (0.0015 to 0.057)      (0.003 to 0.05)
                      ([section])           ([section])

Full                    -0.084                -0.022
service            (-0.174 to 0.005)     (-0.107 to 0.062)

Other                    -0.01                -0.017
service           (-0.018 to -0.003)    (-0.025 to -0.009)
                      ([section])           ([section])

Restaurant             Model 4 Urbanicity: p (95% CI))

type                      CMA                 Non-CMA
([dagger])

Fast-food                0.037                -0.003
service            (0.023 to 0.052)      (-0.025 to 0.019)
                      ([section])

Full                     -0.06                -0.054
service            (-0.11 to -0.011)     (-0.15 to 0.038)
                      ([section])

Other                   -0.016                -0.0006
service           (-0.023 to -0.009)     (-0.013 to 0.012)
                      ([section])

* Survey sampling weights provided by Statistics Canada were
used in all regressions. ([dagger]) Density: outlets per
10,000 forward sortation area population based on the 2006
Canadian Census.

([double dagger]) Individual-level factors controlled for:
demographic characteristics, socio-economic status,
lifestyle and geographic characteristics, dissemination area
level.

([double dagger]) Statistically significant at p<0.05.

Table 4. Association * between BMI and restaurant density
for those living in CMAs, stratified by sex, by restaurant
service type

Restaurant                Model 5 CMA: p (95% CI)
type
([dagger])             Males                 Females

Fast-food         0.032 (0.0009          0.041 (0.01 to
  service      to 0.063) ([section])   0.072) ([section])

Full service      -0.1 (-0.19 to             -0.008
               -0.0018) ([section])      (-0.1 to 0.08)

Other            -0.01 (-0.018 to      -0.024 (-0.034 to
  restaurant    -0.001) ([section])    -0.014) ([section])
  service

* Survey sampling weights provided by Statistics Canada were
used in all regressions.

([dagger]) Density: outlets per 10,000 forward sortation
area population based on 2006 Canadian Census.

([double dagger]) Individual-level factors controlled for:
demographic characteristics, socio-economic status,
lifestyle and geographic characteristics, dissemination area
level.

([section]) Statistically significant at p<0.05.