An exploratory spatial analysis of overweight and obesity in Canada
Introduction
For many years, overweight and obesity have been studied as important risk factors for cardiovascular and other chronic diseases with major implications for individuals as well as populations (National Institute of Health, 2004). Overweight/obesity increase the risk of diabetes and heart disease, contribute to greater risk for hypertension and insulin resistance as well as impact the psychosocial state and quality of life of individuals and families (Raine, 2004). Further, overweight/obesity have been associated with increased prevalence of primary-care visits, exacerbating the load on an already overburdened health-care system (Bertakis and Azari, 2005). According to Katzmarzyk and Janssen (2004), the cost associated with obesity in Canada in 2001 was $4.3 billion, representing 2.2% of total health-care costs.
The problem of excess adiposity was recognised in Canada almost 50 years ago and was the stimulus for the 1953 Canadian Weight-Height Survey (CWHS) (Katzmarzyk, 2002a). The CWHS was the first nationally representative survey of measured stature and body mass in Canada. Since then, data on obesity trends have been informed by a number of different surveys based primarily on self-reported information. Survey data indicate that the national prevalence of obesity in adults increased from 5.6% in 1985 to almost 15% in 1998 (Katzmarzyk, 2002b). More recent self-reported data from the 2005 Canadian Community Health Survey (CCHS) identified that approximately 4 million adults (17.55%) were obese and an additional 8 million adults (33.77%) were overweight.
Few studies have also documented regional differences in the prevalence of obesity in Canada (Katzmarzyk, 2002b). An exception is Katzmarzyk (2002b) who presented surveillance maps for Canadian adults from 1985 to 1998. In 1985, all provinces reported prevalence of obesity less than 10%, while in 1998, all provinces except British Columbia and Quebec reported prevalence of more than 15%. Further, based on the 2005 CCHS, the prevalence of obesity in most of the Atlantic Provinces, Saskatchewan, Nunavut, Yukon and the Northwest Territories was more than 20%.
However, most of these studies have focused on national/provincial levels of aggregation, making it very difficult to illustrate clearly vulnerable regions. A notable exception is a study by Vanasse et al. (2006), which indicates that interest has started shifting from global to local relationships. In particular, Vanasse et al. (2006) assessed the relationship between obesity, leisure-time physical activity and daily fruit and vegetable consumption at the health region level across Canada. Findings revealed that the prevalence of obesity varied substantially between health regions with higher values being strongly associated with low values of physical activity and fruit and vegetable consumption. Further, Vanasse el al. (2006) made use of cartograms to adjust the distortion caused by huge differences in population density between health regions. However, the use of cartograms assumes that observations (i.e. overweight/obesity rates) are independent of each other. Yet, prevalence of overweight/obesity at one health region are likely to be correlated with prevalence in nearby regions, indicating the presence of clusters (Bailey and Gatrell, 1995). Therefore, we felt that it would be worthwhile to examine the overweight/obesity spatial patterns taking into account the overall as well as local variability in the data. In particular, cluster analysis is applied in order to identify areas with higher rates of overweight/obesity as well as generate hypotheses based on the emerging spatial patterns. Further, this study illustrates that cluster analysis provides a great potential to uncover region-specific risk factors of overweight/obesity as well as be a useful tool for public health professionals and policy makers to effectively direct scarce public health resources on vulnerable regions.
Section snippets
Data sources and study design
We conducted a cross-sectional population based ecological-level analysis to assess the spatial patterns in overweight and obesity in Canada. The Public Use Microdata File (PUMF) of the CCHS 3.1 cycle was used to obtain information on the prevalence of overweight/obesity among Canadians. The CCHS is a cross-sectional national survey (n = 132,221) that covers the population aged 12 and older in all provinces and territories, except individuals living on Indian reserves, institutions and some
Results
Results revealed that, age-standardised rates of overweight/obesity were higher in males than females (Fig. 1, Fig. 2, Fig. 3). This difference is seen at the provincial level, as well (Tjepkema, 2005). Also, the three major metropolitan areas of Vancouver, Toronto and Montreal exhibited overweight and obesity rates below the Canadian overall standardised rates of 34.99 and 18.53% respectively (Fig. 1).
Age-standardised rates of overweight differed substantially across health regions ranging
Discussion
This study illustrates the marked variability of overweight/obesity prevalence in Canada at the health region level. In addition, the significant positive spatial autocorrelation identified by Moran's I for both males and females as well as the significant local clusters identified by LISA confirm the presence of spatial heterogeneity (Anselin, 1994) and suggest the importance of developing/refining prevention programmes and health-related policies that are thus far mainly based on assumptions
Study limitations
There are two issues that need to be addressed. First, the overweight/obesity data used were based on self-reported information. However, given that individuals tend to over-report their stature and under-report their weight (Katzmarzyk, 2002a) the prevalences' of overweight/obesity reported in this study should be considered conservative. Second, the spatial patterns of overweight/obesity identified may change depending on the spatial scale used, an issue which is also referred to as the
Conclusions
The findings established in this study strengthen previous studies results' by confirming marked sex-specific regional differences in the prevalence of overweight and obesity in Canada. In addition, the identification of spatial clusters provides additional valuable information on the local geographical variation of overweight/obesity in Canada. Application of Moran's I and LISA verify the effectiveness of spatial data analysis methods and in particular cluster analysis as a helpful tool for
Endnote
The term ‘hot spots’ is widely used in spatial statistics (Anselin, 1994, Bailey and Gatrell, 1995), and has been used in various disciplines for identifying critical locales in a wide range of fields, including seismology (Chulick et al., 2001), epidemiology (Crighton et al., 2007, Tiwari et al., 2006) and criminology (Ratcliffe, 2004). Essentially, ‘hot spots’ simply refers to a significant cluster of homogenous data values.
Conflicts of interest statement
The authors declare that this research forms part of the Canadian Heart Health Surveys Follow-Up Study, an interdisciplinary, multi-site study funded by the Canadian Institute for Health Research (CIHR) and the Heart and Stroke Foundation of Canada.
Acknowledgments
The authors wish to thank Dr. K. Bruce Newbold for his helpful feedback with the analysis of the data. We also extend our appreciation to the Data Liberation Initiative (DLI) at McMaster University.
References (25)
- et al.
Dietary intake and body mass index of adults in 2 Ojibwe communities
J. Am. Diet. Assoc.
(1999) Problems of spatial analysis in geographical epidemiology
Soc. Sci. Med.
(1979)Exploratory spatial data analysis and geographic information systems
- Anselin, L., 1995. Local indicators of spatial association — LISA Geographical Analysis, 27, pp....
- Anselin, L., 2003. GeoDa 0.9 User's Guide, Spatial Analysis Laboratory, University of Illinois, Urbana-Champaign,...
- et al.
Interactive Spatial Data Analysis
(1995) - et al.
Obesity and the Use of Health Care Services
Obes. Res.
(2005) - Case, A. and Menendez, A., 2007. Sex differences in obesity ratss in poor countries: evidence from South Africa NBER...
- et al.
Crustal structure of North America and the adjacent ocean basins
Geol. Soc. Am.
(2001) - et al.
An exploratory spatial analysis of pneumonia and influenza hospitalizations in Ontario by age and gender
Epidemiol. Infect.
(2007)
Social, educational, and psychological correlates of weight status in adolescents
Obes. Res.
Cited by (52)
GIS-Based Study of Dental Accessibility and Caries in 3-Year-Old Japanese Children
2023, International Dental JournalCitation Excerpt :Accordingly, we computed the empirical Bayesian estimation of caries prevalence to provide more stable estimates. To examine whether the overall spatial clustering of the AIR at the neighboring municipalities’ level was random, dispersed, or clustered and whether the values observed at one location correlated with those observed in neighbouring municipalities’ locations, we employed the global Moran's I statistic, which varied between −1 and +1.38 Next, we applied the local indicator of spatial autocorrelation (LISA) to gauge local spatial association.39
Spatial clustering patterns of child weight status in a southeastern US county
2018, Applied GeographyCitation Excerpt :Often referred to as a hot-spot analysis, this clustering test provided an indication of the degree that localized areas represent unexpectedly high or low BMI z-score values compared to the overall, or global BMI z-score average across the sample (Anselin, 1995). Furthermore, this test can identify five categories of various spatial patterns that may be present in the data: Not Significant, High-High, High-Low, Low-High, and Low-Low (Pouliou & Elliott, 2009). This study was particularly interested in the High-High and Low-Low patterns of clustering that represent areas where youth with high and low BMI values, respectively, are surrounded by youth with similar values, indicating areas of geographic concentrations of high or low youth obesity.
Exploring spatial trends in Canadian incidence of hospitalization due to myocardial infarction with additional determinants of health
2016, Public HealthCitation Excerpt :Furthermore, spatial trends have been observed for overweight and obesity rates in Canada that vary across geographic regions administered by various health authorities.5 Considering the link between issues of overweight and obesity with cardiovascular health,6–8 it is reasonable to assume that the incidence of MI might also demonstrate spatial homogeneity across the country that follows, to some extent, the spatial trends observed for obesity and other closely related health issues.4,5 Despite myriad suggestions related to lifestyle changes for improving cardiovascular health, the expense associated with MI incidence presents a persistent drain on available health care resources and personal finances.9,10
Spatial patterns of adolescent drug use
2015, Applied GeographyCitation Excerpt :Although global spatial autocorrelation methods have existed for half a century, it was not until the early 1990s that local methods were developed for identifying clusters (McLafferty, 2008). The local method has been used in a variety of public health research studies ranging from park access and obesity to neighborhood homicides (Messner et al., 1999; Pouliou & Elliott, 2009; Talen & Anselin, 1998). Likewise, Frick and Castro (2013) reported clusters of high tobacco retail density near schools within districts that are social disadvantaged.
Spatiotemporal heterogeneity of malnutrition indicators in children under 5 years of age in Bangladesh, 1999-2011
2018, Public Health Nutrition