Obesity in Canada – Determinants and contributing factors

Determinants and Contributing Factors

Obesity is a complex phenomenon that involves a wide and interactive range of biological, behavioural and societal factors.Footnote 43-46 While genetics play a role, genes do not operate in a vacuum; behaviours and social, cultural and physical environments also make important contributions.Footnote 47

A population health approach looks at patterns of health across different populations and also considers a range of determinants or factors associated with the health outcome. This section discusses current evidence and analysis for a range of behavioural and contextual factors associated with obesity in Canada.

Research suggests that, to be relevant to health in Aboriginal populations, frameworks of health determinants need to address the specificity of the experiences of those populations.Footnote 48–49 Footnote 50 Aboriginal populations have distinct histories, but they share common experiences of colonialism, racism and social exclusion.Footnote 48 Reflecting these histories and a more holistic cultural perspective on health,Footnote 49 for Aboriginal peoples the range of determinants of health may also include factors such as cultural continuity and the relationship to land.Footnote 50 Although it is not the goal of this report to explore these issues at length, the historical experiences of Canada's Aboriginal peoples provide important context in considering the determinants of Aboriginal health, including obesity.

Physical activity

There is considerable evidence of an inverse relation between the prevalence of obesity and leisure-time physical activity (LTPA).Footnote 12 Footnote 46 Footnote 51 Energy expended during non-exercise activity (known as "non-exercise activity thermogenesis" [NEAT])Footnote 52 Footnote 53 includes activities of daily living, occupational or work-related activity, active commuting and incidental movement. Evidence is still emerging, but it has been suggested that the relation between physical activity and health outcomes such as obesity may be moderated by a number of lifestyle factors, including NEAT activities, sedentary behaviours and sleep.Footnote 53

For the most part, physical activity studies in Canada have tended to focus on LTPA.Footnote 54 Many of these studies have relied on self-reported data that may be susceptible to respondent and response bias.Footnote 55 Systematic reviews have suggested that indirect (e.g., questionnaire or diary) and direct (e.g., accelerometry) measures may produce differing estimates of physical activity in adults,Footnote 56 and children and youth.Footnote 57

Available data show that many Canadians get less than the daily recommended amount of physical activity for their age group. The OECD has suggested that, in addition to an obesity epidemic, "there is also a less visible, but no less important, epidemic of 'lack of cardio-respiratory fitness'."Footnote 20 The Canadian Physical Activity Levels Among Youth (CAN PLAY) study estimated that during the 2007-2009 period, 88% of children and youth aged 5 to 19 did not meet the guidelines of Canada's Physical Activity Guide.Footnote 58 In the 2007/08 CCHS, only half (51%) of Canadians aged 12 and over were active or moderately active (analysis not shown here). In the 2007-2009 CHMS, the proportion of adults whose aerobic fitness was categorized as "fair" or "in need of improvement" increased with age, from 32% of males and 20% of females aged 15 to 19 yearsFootnote 26 to 59% of males and 92% of females aged 60 to 69 years.Footnote 2

Sedentary behaviours and screen time

Sedentary behaviours include screen time (i.e., time spent watching television or videos or using a computer), reading, sitting during transit and sedentary hobbies. Being sedentary is often confused with physical inactivity, but the relation between the two is still unclear.Footnote 59 As with physical activity, sedentary behaviour can be measured directly or indirectly, and conducting research can be methodologically challenging.Footnote 60

A high level of screen time is associated with a greater likelihood of being obese for Canadian adultsFootnote 61 and children.Footnote 26 Footnote 62 One study found that overweight and non-overweight boys and girls in Canada did not differ significantly by reported physical activity patterns but did differ by screen time, in that overweight groups were more likely to spend two hours or more in front of a screen daily.Footnote 63

According to the 2009 Report Card on Physical Activity for Children and Youth by Active Healthy Kids Canada, only 19% of children and youth are currently meeting the guideline of less than two hours per day of screen time.Footnote 64 Screen time for both adultsFootnote 65 and childrenFootnote 66 is influenced by a number of demographic and socioeconomic factors, including age, sex, education, household income and urban vs. rural residency. Screen time can also vary by the type of screen time activity.

Diet

Along with physical (in)activity, diet is the most well-studied behavioural factor influencing body weight and overweight and obesity risk. Although much of the available evidence is limited to correlational findings, overall, the balance of the data underscores the importance of healthy eating patterns and access to healthy food as key factors associated with obesity at a population level.

A number of studies have found an association between low consumption of fruits and vegetables, an indicator of a poor diet, and obesity.Footnote 12 Footnote 46 As well, modelling research of Canadian energy intake and expenditure levels from 1976 to 2003 has also shown a strong association between rising obesity prevalence and rising energy consumption, with most of the latter accounted for by seven food commodities (salad oils, wheat flour, soft drinks, shortening, rice, chicken and cheese).Footnote 67

In children and adolescents, familial and environmental factors may be associated with dietary choices and behaviours.Footnote 68–70 For example, snacking or eating dinner while watching television,Footnote 71 Footnote 72 consumption of sugar-sweetened beverages between mealsFootnote 73 and skipping breakfastFootnote 74 have been associated with an increased risk of obesity in children and youth. As well, a study of middle-school-aged children found that a greater frequency of family dinners was associated with less soft drink consumption, more frequent breakfast eating, less concern over high body weight and higher self-efficacy for healthy eating at home and during social times with friends.Footnote 75

More broadly, food insecurity (defined as an income-related problem in accessing food)Footnote 76 during the preschool years has been found to increase the likelihood of overweight later in childhood.Footnote 77 However, a relationship between food insecurity and overweight or obesity has not been shown among adult men, and findings have been inconsistent for adult women.Footnote 76

Socioeconomic status

Analyses of the 2007/08 CCHS suggest that the relation between income and obesity varies by sex (analysis not shown here). Among females, as income increases obesity tends to decrease, a pattern not observed for males. This inverse trend between income and obesity for females and lack of an apparent pattern for males has also been observed among Aboriginal peoples (Figure 9).

Figure 9: Prevalence of Self-Reported Obesity among Aboriginal Peoples by Sex and Income, Ages 18 and Older, 2006

Figure 9: Prevalence of Self-Reported Obesity among Aboriginal Peoples by Sex and Income, Ages 18 and Older, 2006
Figure 9 - Text equivalent

Figure 9 shows the prevalence of self-reported adult obesity among Aboriginal males and females by income category.  Among males: less than $20,000 (25.4%); $20,000 to $39,999 (22.5%); $40,000 to $59,999 (25.6%); $60,000 to $79,999 (27.5%); $80,000 to $99,999 (27.9%); and $100,000 or more (26.3%). Among females: less than $20,000 (26.8%); $20,000 to $39,999 (27.5%); $40,000 to $59,999 (27.5%); $60,000 to $79,999 (23.7%); $80,000 to $99,999 (22.3%); and $100,000 or more (16.3%).

Source: Analysis of the Aboriginal Peoples Survey 2006 Public Use File, Statistics Canada.

Education is another key dimension of socioeconomic status (SES). A generally inverse pattern between education level and obesity prevalence was observed for both men and women in the total Canadian population aged 25 and older (analysis not shown here). Similarly, for the Aboriginal population aged 18 and older (Figure 10), obesity appears less prevalent among men and women with the highest levels of educational attainment.

Figure 10: Prevalence of Self-Reported Obesity among Aboriginal Peoples by Sex and Education, Ages 18 and Older, 2006

Figure 10: Prevalence of Self-Reported Obesity among Aboriginal Peoples by Sex and Education, Ages 18 and Older
Figure 10 - Text equivalent

Figure 10 shows the prevalence of self-reported adult obesity among Aboriginal males and females by level of educational attainment. Among males, some high school (25.3%); high school (25.5%); college/trade diploma (28.3%); and university (21.4%). Among females: some high school (27.2%); high school (25.6%); college/trade diploma (25.2%); and university (17.5%).

Source: Analysis of the Aboriginal Peoples Survey 2006 Public Use File, Statistics Canada.

A study of body weight and occupational prestige reported different patterns for men and women. Among men, after adjusting for age, income and education, no linear associations between occupational prestige and overweight were found. Among women, increasing occupational prestige was associated with lower BMI on average, even after adjusting for age and income. However, this effect was almost eliminated after education had been taken into account, suggesting that, for women, the relation between occupational prestige and BMI is largely attributable to education.Footnote 78

Community-level factors

Analyses have shown that indicators of area- or neighbourhood-level SES are correlated with obesity in adults,Footnote 79 and children and youth.Footnote 70 Footnote 80 Footnote 81 New analysis of data from the 2005, 2007 and 2008 CCHS looked at disparities in obesity by SES in Canada's Census Metropolitan Areas (CMAs). In most CMAs, obesity was more prevalent in the most socioeconomically deprived areas than in the least deprived (Figure 11). In Halifax, for example, 25.5% of people in the lowest SES areas were obese compared with 11.2% of people in the highest SES areas. However, in some CMAs, no significant disparities were found. Results and detailed maps identifying low SES areas for all CMAs are available on the CIHI website.Footnote 82

Figure 11: Prevalence of Self-Reported Obesity by Area-Level SES in Select Census Metropolitan Areas, 2005-2008

Figure 11: Prevalence of Self-Reported Obesity by Area-Level SES in Select Census Metropolitan Areas, 2005-2008
Figure 11 - Text equivalent

Figure 11 shows the prevalence of self-reported obesity by area-level socioeconomic status (SES; high versus low) in selected census metropolitan areas in 2005-2008: St. John’s, NL (15.8% in high-SES areas versus 23.1% in low-SES areas); Halifax, NS (11.2% versus 25.5%); Québec, QC (8.9% versus 12.4%); Montreal, QC (10.3% versus 15.8%); Ottawa-Gatineau, ON/QC (10.1% versus 15.5%); Toronto, ON (11.3% versus 13.9%); Winnipeg, MB (11.7% versus 16.5%); Regina, SK (14.0% versus 26.0%); Calgary, AB (8.3% versus 16.5%); Edmonton, AB (16.2% versus 17.8%); Vancouver, BC (8.0% versus 8.3%); and Victoria, BC (10.2% versus 14.1%).

Note: * Significantly different from High SES estimate at p<0.05.

Source: Analysis of the 2005 and 2007/08 Canadian Community Health Surveys, Statistics Canada.

One avenue through which neighbourhood physical and sociocultural characteristics may influence obesity risk is their impact on the availability and accessibility of physical activity equipment, facilities or programs, though the direction and extent of influence may vary by age.Footnote 83 Other research has shown that the impacts for children vary by urban and rural residence: while access to recreational facilities and shops with modestly priced healthy foods was associated with less obesity, the former was particularly important to the activity level and body weight of children in rural areas, whereas the latter was particularly influential in the diet and body weights of children from urban areas.Footnote 69

Another possible avenue of influence is through access to food retail outlets.Footnote 47 Footnote 84 A study in Edmonton, for example, showed that the odds of being obese increased with the concentration of convenience stores and fast-food outlets in the neighbourhood, regardless of covariates such as neighbourhood SES, age, sex and education.Footnote 85 However, the evidence of a relation between obesity and the community food environment is mixed.Footnote 86

Community consumption of traditional foods has been shown to be associated with lower rates of obesity among First Nations children. In the 2002/03 RHS, compared with children in large First Nations communities (i.e., 1,500 or more residents), those who lived in small communities of less than 300 were more likely to consume traditional foods and less likely to be obese (the prevalence of obesity being 25.7% in small communities versus 44.2% in large communities). Among First Nations adults and youth, the association between community size and consumption of traditional foods remained but did not appear to be related to BMI.Footnote 28

Box 4. Adjusted Population Attributable Risk and Population Impact Number

Adjusted Population Attributable Risk (PARadj)

Population attributable risk (PAR) is a measure of the theoretical reduction in disease incidence that would be observed in a population if a given risk factor were entirely eliminated, after controlling for other factors. It is calculated by multiplying the relative risk (RR) of the disease associated with that risk factor by the proportion of the population exposed to the risk factor. An adjusted PAR (PARadj) uses an RR that is adjusted for other factors, such as social determinants or health behaviours.

Population Impact Number (PIN)

A PIN is a measure of the number of cases of a certain disease or condition in a population that may be attributed to a given risk factor, after controlling for other factors, and reflects the potential reduction in the number of people in that population with the disease if that risk factor were entirely eliminated. It is calculated by multiplying the PARadj by the proportion exposed and by the number of people in the population.

For additional details about the methodology and descriptive estimates of the risk factors, see Appendix 3.

Contribution of multiple risk factors to obesity

For this report, adjusted population attributable risks (PARsadj) were calculated to estimate the proportion of overweight and obesity in the population that is attributable to specific demographic, social and behavioural risk factors, while taking into account (i.e., adjusting for) their correlation with other factors.

Two types of risk factor for overweight and obesity were included in this analysis:

  • Social determinants: immigrant and visible minority status, household income (low, middle or high), urban vs. rural residence, and marital status; and
  • Health behaviours: LTPA, smoking status, fruit and vegetable consumption, and alcohol consumption.

Figure 12 shows the PARadj of obesity associated with each of the six social determinant and four health behaviour risk factors. After adjustment for other factors including age, income, rural residence, and alcohol and cigarette use, low levels of LTPA emerged as having the strongest association with obesity at the population level for both men and women. But the analysis also found that LTPA was more strongly associated with obesity among women than it was among men (other types of physical activity were not included in this analysis).

Similarly, living on a low income was associated with obesity among women (again, after controlling for the other factors) but was not associated with obesity among men. In contrast, an association was found between consuming less than five fruits and vegetables daily and obesity for both men (7.9%) and women (3.9%).

Figure 12: Population Attributable Risk of Self-Reported Obesity, by Risk Factor and Sex, Ages 18 Years and Older, Canada

Figure 12: Population Attributable Risk of Self-Reported Obesity, by Risk Factor and Sex, Ages 18 Years and Older, Canada
Figure 12 - Text equivalent

Figure 12 shows the population attributable risk (PAR) of self-reported adult obesity among females and males, by risk factor. For females, the PAR of self-reported obesity associated with: immigrant status (-2.5%); visible minority status (-6.8%); lowest income quintile (4.3%); highest income quintile (-4.6%); rural residence (2.1%); single status (-0.9%); low physical activity (21.6%); being a smoker (-4.8%); low fruit and vegetable consumption (3.9%); and high alcohol consumption (-4.0%). For males, the PAR of self-reported obesity associated with: immigrant status (-4.7%); visible minority status (-6.5%); lowest income quintile (-0.5%); highest income quintile (-0.6%); rural residence (1.5%); single status (-9.0%); low physical activity (11.1%); being a smoker (-8.5%); low fruit and vegetable consumption (7.9%); and high alcohol consumption (0.2%). 95% confidence intervals are shown for each bar.

Notes: Error bars represent 95% confidence intervals based on bootstrap variance estimates.

Source: R. Hawes and P. Stewart, unpublished manuscript prepared for the Public Health Agency of Canada; based on analysis of pooled 2000/01, 2003 and 2005 Canadian Community Health Surveys, Statistics Canada.

PARadj can also be used to calculate the population impact number (PIN), the theoretical number of cases of overweight or obesity in a population that may be attributed to a specific risk factor, after taking into account the other risk factors in the study. For this analysis, three categories of excess weight were analyzed separately for men and women: overweight I (BMI = 25.0-27.4 kg/m2), overweight II (BMI = 27.5-29.9 kg/m2) and obesity (BMI > 30 kg/m2).

Figures 13 and 14 show the PINs obtained from the analysis for men and women, respectively. The equivalent of 405,000 cases of male obesity and 646,000 cases of female obesity could be averted if all individuals in the population attained high levels of physical activity, as measured in this study; this is consistent with the large PARadj values for low physical activity shown in Figure 12. Similarly, eliminating the consumption of a poor-quality diet, as measured by low fruit and vegetable consumption, may result in 265,000 fewer men and 97,000 fewer women being obese.

These figures also point to the importance of gender as a mediating factor. For example, whereas heavy alcohol consumption was associated with 190,000 cases of overweight among men, it was not associated with an increase in the number of obese men and did not substantially influence the number of overweight or obese women in Canada. Also, the findings suggest that shifting the risk profile of low-income people to that of high-income people could result in about 114,000 fewer women in the population being classified as overweight I, 158,000 fewer women as overweight II and 119,000 fewer women as obese, but may not be associated with changes in overweight or obesity among men.

Figure 13: Population Impact Number of Self-Reported Overweight and Obesity Among Males, by Risk Factor and Body Mass Index Category, Ages 18 Years and Older, Canada

Figure 13: Self-Reported Overweight and Obesity Among Males, by Risk Factor and BMI Category, Ages 18 Years and Older
Figure 13 - Text equivalent

Figure 13 shows the population impact number (PIN) of self-reported adult overweight and obesity among males in different body mass index (BMI) categories, by risk factor. For the Male overweight I BMI category, the PIN associated with: immigrant status (-6,525); visible minority status (-184,989); lowest income quintile (-100,428); highest income quintile (35,002); rural residence (43,716); single status (-387,945); low physical activity (-47,948); being a smoker (-246,837); low fruit and vegetable consumption (109,430); and high alcohol consumption (84,254). For the Male overweight II BMI category, the PIN associated with: immigrant status (-7,881); visible minority status (-276,892); lowest income quintile (-108,106); highest income quintile (11,137); rural residence (61,068); single status (-474,305); low physical activity (-5,090); being a smoker (-317,101); low fruit and vegetable consumption (190,327); and high alcohol consumption (105,402).  For the Male obese BMI category, the PIN associated with: immigrant status (-181,730); visible minority status (-345,072); lowest income quintile (-16,581); highest income quintile (-21,109); rural residence (53,005); single status (-365,423); low physical activity (404,701); being a smoker (-351,416); low fruit and vegetable consumption (265,188); and high alcohol consumption (6,594).

Notes: Error bars represent 95% confidence intervals based on bootstrap variance estimates.

Source: R. Hawes and P. Stewart, unpublished manuscript prepared for the Public Health Agency of Canada; based on analysis of pooled 2000/01, 2003 and 2005 Canadian Community Health Surveys, Statistics Canada.

Figure 14. Population Impact Number of Self-Reported Overweight and Obesity Among Females by Risk Factor and Body Mass Index Category, Ages 18 Years and Older, Canada

Figure 14. PIN of Self-Reported Overweight and Obesity Among Females by Risk Factor and BMI Category, Ages 18 Years and Older
Figure 14 - Text equivalent

Figure 14 shows the population impact number (PIN) of self-reported adult overweight and obesity among females in body mass index (BMI) categories, by risk factor.  For the Female overweight I BMI category, the PIN associated with: immigrant status (37,413); visible minority status (data not available); lowest income quintile (113,847); highest income quintile (-61,184); rural residence (42,379); single status (-198,253); low physical activity (228,004); being a smoker (-87,665); low fruit and vegetable consumption (19,994); and high alcohol consumption (-35,869). For the Female overweight II BMI category, the PIN associated with: immigrant status (12,406); visible minority status (-186,001); lowest income quintile (158,099); highest income quintile (-84,753); rural residence (65,553); single status (-227,353); low physical activity (352,899); being a smoker (-85,585); low fruit and vegetable consumption (29,718); and high alcohol consumption (-67,690). For the Female obese BMI category, the PIN associated with: immigrant status (-66,658); visible minority status (-284,847); lowest income quintile (119,199); highest income quintile (-153,608); rural residence (55,717); single status (-21,428); low physical activity (645,940); being a smoker (-138,075); low fruit and vegetable consumption (97,336); and high alcohol consumption (-140,011).

Notes: Error bars represent 95% confidence intervals based on bootstrap variance estimates.

Source: R. Hawes and P. Stewart, unpublished manuscript prepared for the Public Health Agency of Canada; based on analysis of pooled 2000/01, 2003 and 2005 Canadian Community Health Surveys, Statistics Canada.

Box 5. Cautionary Note on Interpreting PARs and PINs

Population attributable risks (PARs) and population impact numbers (PINs) are useful measures for communicating characteristics of factors that may be associated with the prevalence of a disease or condition at a population level. However, caution should be used when interpreting their results. For example, PARs are non-additive, so individual PAR values for several risk factors cannot be summed together to derive an estimate of "total attributable risk" for the disease or condition of interest. This is because risk factors often cluster and influence one another, particularly in complex health issues such as obesity.

Another issue concerns the interpretation of PARs and PINs for non-modifiable risk factors. In general, where causality is known, these measures can be seen as reflecting the extent of the population burden of a disease or condition (e.g., obesity) that could be theoretically "eliminated" if all individuals in the exposed/target group (e.g., low physical activity) were converted to the non-exposed/referent group ("adequate" physical activity). Such an interpretation, which can help to inform decisions in public health settings about the modifiable risk factors on which to focus limited resources and efforts, is inappropriate when considering non-modifiable risk factors (e.g., immigrant status, urban vs. rural residence). However, the inclusion of such risk factors in PAR (and PIN) analyses can still be of value for informing public health action, as they can help to clarify which groups appear to be at higher or lower risk.

These estimates are theoretical and intended to illustrate in clear population terms the potential magnitude of change to overweight and obesity arising from various behavioural and social factors. To be valid, PAR estimates require an assumption of a cause-and-effect relation between the risk factor and outcome of interest. Such assumptions were made for the purposes of these analyses. This necessarily oversimplifies the complex relations between obesity and its various drivers, particularly with respect to the more distal, or indirect, social determinants of obesity. However, the more pathologically distal factors, such as income, rural residence and minority status, continue to affect male and female overweight and obesity even after controlling for more proximal, or direct, determinants, like the health behaviours analyzed above. This suggests that a) social factors may have a measurable and direct effect on overweight and obesity, and/or b) contextual factors affect overweight and obesity through other, more proximal, determinants not investigated in the CCHS.

In a recent U.K. study that used a similar analytical approach to explore the potential population impact of several neighbourhood-level factors on physical activity, the authors noted that “in practice, given the paucity of community-based evaluations, policy-makers often rely on cause-effect relationships to be assumed to some degree” and that their analysis “merely applied a population perspective to such interpretation.” Nevertheless, they underscored that their results, which assume a cause-effect relation, should be interpreted with caution.Footnote 87 Similar discretion should be used in considering the findings presented above.

Analyses such as the multifactorial research summarized in this chapter are providing new insights into the complex ways in which factors interact and contribute to obesity. However, there is still much to learn, for example:

  • the effects of biological or genetic influences and pre- and post-natal factors, including birth weight and breastfeeding;
  • how factors might differ for different populations, cultures and ethnic groups;
  • the contribution of incidental, life-style-embedded and occupational activities, as well as sedentary behaviours, to physical activity and the risk of obesity; and
  • the effect of socioeconomic and environmental factors such as food security, access to stores and recreational facilities, food supply factors, as well as the built environment.

In the future, further refinement and use of techniques incorporating multiple risk factors (such as PARadj and PIN) may be helpful in gaining insights into the distribution of obesity, as well as indicating opportunities for health promotion and prevention.

Key points

  • Obesity is a complex phenomenon with a wide range of genetic, lifestyle, social, cultural and environmental factors contributing to variations in its prevalence.
  • The association between income and obesity appears to be sex-specific, with an inverse association observed for females in the total population as well as Aboriginal populations but no clear pattern for males.
  • Of the factors considered and currently measured through the CCHS, being inactive emerged as having the strongest association with obesity at the population level for both men and women.
  • An estimated 405,000 cases of male obesity and 646,000 cases of female obesity could potentially be altered or averted if inactive populations became active.
  • Distal or indirect factors, such as income, rural residence and minority status, continue to affect male and female obesity even after controlling for more proximal, or direct, health behaviours such as inactivity.
  • Two population health measures – the PARadj and PIN – provide new perspectives on obesity and the potential contribution of specific factors to obesity prevalence, and they may be one consideration in setting priorities for the prevention and management of obesity.
  • More research into the determinants of obesity is needed, particularly multifactorial research that looks at biological, environmental, socioeconomic and lifestyle factors and how they interact.
  • More research is needed to understand the determinants of obesity – both direct and more indirect – that may be specific to Aboriginal peoples and communities.
  • A limitation of using data and analysis to inform policy is that food-related factors (access to healthy foods and food outlets, consumption of traditional diets, caloric density, marketing of foods and beverages high in sugar and fat to children, and portion sizes) have not been considered in the analysis.

Page details

Date modified: