## Abstract

Background The prevalence of diabetes has been steadily increasing in Western countries. We investigated the impact of socioeconomic status (SES) on the prevalence of self-reported diabetes, and its differences between genders.

Methods Data for this investigation were derived from the second cycle of the National Population Health Survey conducted in 1996–1997. A total of 39 021 subjects (17 730 males and 21 291 females) ≥40 years of age who answered the question about diabetes were included in the present analysis. Educational attainment and income adequacy were used as indicators of SES. Multiple logistic regression models were constructed for men and women separately to assess the effects of SES on the prevalence of diabetes after adjustment for age, area of residence, body mass index, and physical activity.

Results and The prevalence of diabetes was 6.6% among men and 5.5% among women. The

Conclusions prevalence increased with decreasing income category and educational attainment in both genders. The odds ratios for income and education in relation to diabetes after adjustment remained significant in women, but attained unity in men. Canadian women ≥40 years of age of low SES have a relatively high prevalence of diabetes, independent of age, area of residence, obesity, and physical inactivity.

The prevalence of diabetes has been steadily increasing in Western countries.1 In Canada, diabetes is considered to be one of the main preventable diseases associated with premature deaths.2 Socioeconomic status (SES), which plays an important role in health care and disease prevention, is a complex indicator of health services access, knowledge of health promotion, willingness to seek treatment, and lifestyle behaviour.

Educational attainment and income adequacy are important indicators of SES. Although studies from developing countries have demonstrated a positive association between SES and diabetes,3,4 low SES tends to be associated with a high prevalence of this disease in developed countries.5–7 These studies support the notion that in populations that have not gone through, or are going through, the epidemiological transition diabetes is positively associated with SES, but that post epidemiological transition it has a negative association.

There is evidence for a negative association between SES and the prevalence of diabetes in Canada. Millar et al.8 found that the prevalence of diabetes tended to be higher among less-educated men and women. An analysis based on the first cycle of the National Population Health Survey (NPHS) suggested that the prevalence of diabetes was associated with low income.9 Since there is lack of evidence on gender-related differencesin the association between SES and diabetes, this issue was addressed here.

## Materials and Methods

The present analysis was based on data from the second cycle of the NPHS conducted by Statistics Canada in 1996–1997. The NPHS is designed to collect information related to the health of the Canadian population. The first cycle of data collection was carried out in 1994–1995, and continued every second year thereafter. Compared with the first cycle, the second cycle included significantly more study subjects.

Subjects were sampled from each province excluding those living in Indian reserves, Canadian military bases, and some remote areas in Quebec and Ontario. The NPHS used a multi-stage stratified sampling design to select households. One person from each household was randomly selected for the in-depth health survey. A total of 81 804 people participated in the survey, with a response rate of 82.6% (73 402 aged ≥12 and 8402 <12).10

The survey was conducted by a telephone interview, and included items on health status and health determinants such as SES, physical activity, body weight, height, and use of health services. The present analysis was based on 39 021 subjects (17 730 men and 21 291 women) who were ≥40 years old.

Respondents who answered the following question with an affirmative response were considered as diabetic patients: ‘Do you have diabetes diagnosed by a health professional?’ Socioeconomic status indicators included educational attainment and income adequacy. Education was grouped into three levels based on the years of education individual gains: less than secondary school education, which indicates not finishing 12 years’ basic education, secondary school education completed, and post-secondary school education, which indicates gaining more after 12 years’ basic education. Income adequacy was classified into low-, middle-, and high-income groups on the basis of total household income and the number of household members (Table 1).

Potential confounding factors included in this analysis were age, area of residence, body mass index (BMI), and physical activity. Body weight and height were recorded for those under 65 years by self reporting, and BMI was calculated from the equation

; subjects were grouped into <30.0 or ≥30.0 categories. Physical activity was grouped as ‘active‘, ‘moderate‘, or ‘inactive’ based on leisure time energy expenditure.10

The association between SES and diabetes was examined among men and women separately. We calculated the prevalence of diabetes and associated 95% CI stratified by SES and other risk factors. Statistical significance between SES groups was determined if there was no overlap for the 95% CI of the prevalences. Logistic regression was used to evaluate associations between indicators of SES and the prevalence of diabetes after adjustment for other factors. The observations with missing residence area and education were excluded from the analysis, but those with missing income adequacy and BMI were included and categorized into separate groups considering their substantial amount. Following Chen et al.,11 we used the Rao-Wu bootstrap method for variance estimation in order to take into account the complex survey design. All analysis was conducted using SAS 8.

## Results

The prevalence of diabetes was higher among men than women, 6.6% (95% CI: 5.8–7.5%) versus 5.1% (95% CI: 4.6–5.7%). The prevalence increased with decreasing income and education (Table 2); however, the difference in the prevalence among SES levels was significant only in women.

Both men and women <65 years old with BMI ≥30 kg/m2 had a significantly higher prevalence of type 2 diabetes compared with those with a BMI <30 kg/m2. Specifically, the prevalence of diabetes was 9.2% (95% CI: 6.6–11.8%) among men with BMI ≥30 kg/m2 versus 3.9% (95% CI: 3.0–4.7%) for men with BMI <30 kg/m2, and 9.7% (95% CI: 7.1–12.3%) versus 2.4% (95% CI: 1.7–3.0%) for women. Inactive women had a higher prevalence compared with active women, 5.8% (95% CI: 5.0–6.6%) versus 3.3% (95% CI: 2.4–4.3%). Exercise was not significantly associated with diabetes in men.

The influence of age and area of residence on diabetes has been reported previously.12 Younger people tend to have higher income and education levels. The average age in years was 65, 63, and 52 for the low-, middle-, and high-income groups, and 65, 55, and 52 for the low-, middle-, and high-education groups. The BMI data were only collected for those under 65 years old, and the average BMI was similar across the income and education levels. In addition, low income adequacy was associated with female sex, and both low income adequacy and low education were associated with rural living, and sedentary lifestyle (Table 3).

Residual confounding effect might exit when subjects were grouped into four BMI categories. Body mass index as a continuous variable was further adjusted for the association between SES and diabetes among subjects 45–64 years of age. The information on body weight and height was not collected for those ≥65 years. Low income was significantly associated with the prevalence of diabetes in women (OR = 2.36, 95% CI: 1.35–4.14), but not in men. Women with lower education tended to have a higher risk of diabetes, but no such trend was observed in men.

## Discussion

Cross-sectional data from a representative sample of the Canadian population suggest that low SES is associated with a higher prevalence of self-reported diabetes among individuals ≥40 years of age, with the elevation in risk being greater in women than in men. As indicated below, it is believed that self-reported diabetes in this analysis is mainly type 2 diabetes.

The World Health Organization (WHO) and the American Diabetes Association (ADA) have clarified the diagnosis and classification of type 1 and type 2 diabetes.13,14 Unfortunately, implementation of these diagnostic criteria is impractical in large-scale epidemiological studies. As in most population-based studies, the determination of diabetes in the present analysis was based on self-reported physician-diagnosed diabetes. Type 1 diabetes occurs primarily among young people <30 years.6,15 It is reasonable to assume that self-reported diabetes for people ≥40 years of age is mainly type 2 diabetes. One study has indicated that the reliability of self-reported diabetes was high, with a Kappa of 0.85 in Canada.16

There is evidence of a negative association between SES and the prevalence of type 2 diabetes from previous studies. Connolly et al.6 found that people living in the most deprived areas in a health district of the UK had the highest prevalence of type 2 diabetes. Several studies conducted in the US and other developed countries found a higher prevalence of the disease among subjects who were less educated, of a lower income, or unemployed.5,7 Using data from the 1994–1995 NPHS, James et al.9 found that people of both genders who earned less demonstrated a higher prevalence of diabetes without adjustment for potential confounders.

Educational attainment has been widely used as an indicator of SES because of its relation to income, occupation, and social prestige. Education also reflects knowledge of health issues, willingness to seek health information, and healthy lifestyle behaviours. Education is most strongly and consistently associated with the ability to recall risk factors for cardiovascular disease.17 A study conducted in The Netherlands showed that diabetic subjects with low education utilized fewer services relevant to diabetes care.18 Better-educated Canadian adults tend to have fewer risk factors for cardiovascular disease, most of which also influence diabetes.8

Income adequacy reflects the economic status of a household. Having a higher income means having access to goods and services of greater monetary value, with concomitant health benefits. The consumption of more expensive foods may lead to reduced intake of saturated fat; and similarly, more affluent individuals may enjoy regular physical exercise as a consequence of membership of fitness clubs.

To date, gender-related differences in the association between SES and diabetes have remained unclear. In a study of 353 men and 329 women 35–76 years of age, Unwin et al.19 found a relationship between glucose intolerance and manual social class in women, independent of age, BMI, and waist-hip ratio. A weaker non-significant relationship was also reported in men. A study of 1288 Mexican Americans and 929 Anglos living in three San Antonio neighbourhoods showed that the risk of the onset of diabetes fell with rising SES in both Mexican American men and women, although the decline was less pronounced in men.20

The reasons for these gender-related differences in the association between the risk of diabetes and SES need to be explored. The influence of SES on population health is complex. Men and women in different social classes may demonstrate different perception of health, health behaviours, and lifestyles. In the present analysis, obesity was strongly associated with income, education, and type 2 diabetes. It is well known that low SES is associated with obesity, particularly among women.21 Physical inactivity is another predictor of type 2 diabetes. Well-educated individuals are more likely to engage in physical activities.22 Whereas Ford et al. found that women of higher SES were significantly more active than those of lower SES, this difference was not apparent in men.23

Diet has long been considered as a possible cause of diabetes.24 However, the impact of SES on the dietary structure is complex. Kinsey25 has stated that people of all socioeconomic groups are spending a smaller per cent of their disposable income on food and that the continuous introduction of affordable new healthy foods makes it difficult to differentiate socioeconomic groups in the US by nutritional status. However, studies conducted in Canada showed that lower household income was inversely associated with fat intake and smoking,26 and that certain dietary habits, such as consumption of ‘junk food‘, were most closely related to fat intake among low-income people.27 Differences in dietary structure, including fat intake, smoking and alcohol consumption, still exist among socioeconomic groups defined by income and education. However, there is a lack of evidence for gender differences in dietary fat intake among people in different income groups.

Because of the cross-sectional nature of the NPHS, it is difficult to determine if there is a causal relationship between SES and diabetes. As argued above, low SES may increase the risk of diabetes, although diabetes may also result in low SES level. Diabetic patients may be more likely to be unemployed because of their health status, and therefore receive less income. If this is in fact the case, the effect of low SES on diabetes may be overestimated. On the other hand, Kraut et al. found that, based on a prospective study in Manitoba, Canada, although diabetic individuals with complications were twice as likely to be unemployed than non-diabetic individuals, the difference was not evident for diabetic individuals without complications.28 People with severe long-term chronic conditions were not likely included in this survey. It is reasonable to assume that the percentage having complications for the diabetic patients in this analysis was low. Furthermore, the influence of low SES on the onset of diabetes in deprived population groups may be underestimated because people of low SES are less likely to have routine health examinations, including blood or urine glucose tests.6

In this study, there is a lack of measures of central obesity, and both body weight and height were not objectively measured but self-reported. There is a tendency that men over-report their height and women under-report their weight. Such a reporting bias, however, is small and non-differential to disease outcomes, and has little impact on association estimation.29

In summary, we found that Canadian women ≥40 years of age of low SES had a higher prevalence of type 2 diabetes, independent of age, area of residence, obesity, and physical inactivity. This relationship was weaker in men and accounted for the confounding factors stated above. The definitive reasons for this gender difference remain unclear, and require further investigation.

KEY MESSAGES

• The relationship between socioeconomic status and diabetes was examined among 39 021 Canadian adults (17 730 males and 21 291 females) ≥40 years old.

• The prevalence of diabetes was slightly higher in men than in women, 6.6% versus 5.5% in Canada.

• The prevalence increased with decreasing income.

Table 1

Definition of income adequacy. National Population Health Survey, 1996–1998

Income adequacy Household size
Low income
<$10 000 1–4 <$15 000 ≥5
$10 000–$14 999 1 or 2
$10 000–$19 999 3 or 4
$15 000–$29 999 ≥5
Middle income
$15 000–$29 999 1 or 2
$20 000–$39 999 3 or 4
$30 000–$59 999 ≥5
High income
$30 000–$59 999 1 or 2
$40 000–$79 999 3 or 4
$60 000–$79 999 ≥5
≥$60000 1 or 2 ≥$80 000 ≥3
Income adequacy Household size
Low income
<$10 000 1–4 <$15 000 ≥5
$10 000–$14 999 1 or 2
$10 000–$19 999 3 or 4
$15 000–$29 999 ≥5
Middle income
$15 000–$29 999 1 or 2
$20 000–$39 999 3 or 4
$30 000–$59 999 ≥5
High income
$30 000–$59 999 1 or 2
$40 000–$79 999 3 or 4
$60 000–$79 999 ≥5
≥$60000 1 or 2 ≥$80 000 ≥3
Table 2

The prevalence (%) and 95% CI for self-reported diabetes according to socioeconomic status and other potential risk factors among objectives aged ≥40, based on the National Population Health Survey, 1996–1998

Men Women
No. Cases %a 95% CIb No. Cases 95% CI
a The prevalence estimates were weighted to the general population.
b Bootstrap estimate of 95% CI.
Low 1798 160 7.4 (4.9–10.0) 3570 350 9.0 (7.1–10.8)
Middle 4083 369 8.8 (7.1,10.6) 5172 355 6.2 (4.9–7.6)
High 7951 416 5.7 (4.4–7.0) 7466 266 2.8 (2.1–3.5)
Unknown 3898 243 5.5 (4.5–6.6) 5083 289 5.6 (4.3–6.8)
Education level
Less than secondary school 5808 517 8.8 (6.9–10.7) 7181 626 8.5 (7.1–9.9)
Secondary school 2757 161 6.2 (4.4–7.9) 3843 188 4.5 (3.2–5.7)
Post-secondary school 8858 488 5.6 (4.5–6.7) 9979 431 3.3 (2.6–3.9)
Unknown 307 22 4.5 (2.4–6.7) 288 15 3.7 (1.1–6.2)
Body mass index (m/kg2)
<30 9907 373 3.9 (3.0–4.7) 10 319 279 2.4 (1.7–3.0)
≥30 2289 214 9.2 (6.6–11.8) 2085 218 9.7 (7.1–12.3)
Unknown 5534 601 12.0 (9.8–14.1) 8887 763 8.4 (7.2–9.6)
Physical activity
Active 2992 170 5.4 (3.7–7.1) 2766 135 3.3 (2.4–4.3)
Moderate 3689 188 6.0 (3.9–8.2) 4454 187 3.9 (2.5–5.3)
Inactive 10 226 742 7.1 (5.9–8.2) 13 594 897 5.8 (5.0–6.6)
Unknown 823 88 8.8 (5.3–12.3) 477 41 8.6 (4.5–12.7)
Men Women
No. Cases %a 95% CIb No. Cases 95% CI
a The prevalence estimates were weighted to the general population.
b Bootstrap estimate of 95% CI.
Low 1798 160 7.4 (4.9–10.0) 3570 350 9.0 (7.1–10.8)
Middle 4083 369 8.8 (7.1,10.6) 5172 355 6.2 (4.9–7.6)
High 7951 416 5.7 (4.4–7.0) 7466 266 2.8 (2.1–3.5)
Unknown 3898 243 5.5 (4.5–6.6) 5083 289 5.6 (4.3–6.8)
Education level
Less than secondary school 5808 517 8.8 (6.9–10.7) 7181 626 8.5 (7.1–9.9)
Secondary school 2757 161 6.2 (4.4–7.9) 3843 188 4.5 (3.2–5.7)
Post-secondary school 8858 488 5.6 (4.5–6.7) 9979 431 3.3 (2.6–3.9)
Unknown 307 22 4.5 (2.4–6.7) 288 15 3.7 (1.1–6.2)
Body mass index (m/kg2)
<30 9907 373 3.9 (3.0–4.7) 10 319 279 2.4 (1.7–3.0)
≥30 2289 214 9.2 (6.6–11.8) 2085 218 9.7 (7.1–12.3)
Unknown 5534 601 12.0 (9.8–14.1) 8887 763 8.4 (7.2–9.6)
Physical activity
Active 2992 170 5.4 (3.7–7.1) 2766 135 3.3 (2.4–4.3)
Moderate 3689 188 6.0 (3.9–8.2) 4454 187 3.9 (2.5–5.3)
Inactive 10 226 742 7.1 (5.9–8.2) 13 594 897 5.8 (5.0–6.6)
Unknown 823 88 8.8 (5.3–12.3) 477 41 8.6 (4.5–12.7)
Table 3

The distribution of potential risk factors of diabetes by socialeconomic status, based on the National Population Health Survey, 1996–1997

Income adequacy Education level
Low (5368) % Middle (9255) %  High (15 417) % Unknown (8981) % Less than secondary school (12 989) % Some secondary school (6600) % Post-secondary school (18 837) %
Gender
Male 33.5 44.1 51.6 43.4 44.7 41.8 47.0
Female 66.5 55.9 48.4 56.6 55.3 58.2 53.0
Residence area
Rural 26.5 25.0 21.1 21.4 28.9 21.1 19.4
Urban 73.4 75.0 78.8 78.6 71.1 78.9 80.6
Unknown 0.1 0.0 0.1 0.0 0.0 0.0 0.0
Physical activity
Active 12.6 13.2 17.9 12.3 11.1 15.7 17.1
Moderate 17.8 19.9 23.6 19.0 16.9 21.4 23.7
Inactive 67.4 63.1 56.1 63.5 67.1 59.9 57.2
Unknown 2.2 3.8 2.4 5.2 4.9 3.0 2.0
Income adequacy Education level
Low (5368) % Middle (9255) %  High (15 417) % Unknown (8981) % Less than secondary school (12 989) % Some secondary school (6600) % Post-secondary school (18 837) %
Gender
Male 33.5 44.1 51.6 43.4 44.7 41.8 47.0
Female 66.5 55.9 48.4 56.6 55.3 58.2 53.0
Residence area
Rural 26.5 25.0 21.1 21.4 28.9 21.1 19.4
Urban 73.4 75.0 78.8 78.6 71.1 78.9 80.6
Unknown 0.1 0.0 0.1 0.0 0.0 0.0 0.0
Physical activity
Active 12.6 13.2 17.9 12.3 11.1 15.7 17.1
Moderate 17.8 19.9 23.6 19.0 16.9 21.4 23.7
Inactive 67.4 63.1 56.1 63.5 67.1 59.9 57.2
Unknown 2.2 3.8 2.4 5.2 4.9 3.0 2.0
Table 4

Unadjusted and adjusteda odds ratios for the prevalence of self-reported diabetes in relation to socioeconomic status in men and women, based on the National Population Health Survey, 1996–1997

Men Women
bOdds ratio (95% bootstrap CI).
cAdjusted for age, area of residence, BMI category, and physical activity.
Low 1.3 (0.8–2.1) 1.0 (0.6–1.6) 1.0 (0.6–1.6) 3.4 (2.4–4.9) 2.1 (1.5–3.0) 2.1 (1.4–3.0)
Middle 1.6 (1.2–2.2) 1.1 (0.8–1.6) 1.1 (0.8–1.6) 2.3 (1.6–3.2) 1.7 (1.2–2.3) 1.6 (1.2–2.2)
High 1.0 1.0 1.0 1.0 1.0 1.0
Unknown 1.0 (0.7–1.3) 0.8 (0.6–1.1) 0.8 (0.6–1.1) 2.0 (1.4–2.9) 1.5 (1.1–2.3) 1.5 (1.0–2.1)
Education level
Less than secondary school 1.6 (1.2–2.2) 1.1 (0.7–1.6) 1.0 (0.7–1.6) 2.8 (2.1–3.7) 1.8 (1.3–2.4) 1.7 (1.3–2.3)
Some secondary school 1.1 (0.8–1.6) 1.0 (0.7–1.4) 1.0 (0.7–1.4) 1.4 (0.9–2.0) 1.2 (0.8–1.8) 1.2 (0.8–1.7)
Post-secondary school 1.0 1.0 1.0 1.0 1.0 1.0
Men Women
bOdds ratio (95% bootstrap CI).
cAdjusted for age, area of residence, BMI category, and physical activity.
Low 1.3 (0.8–2.1) 1.0 (0.6–1.6) 1.0 (0.6–1.6) 3.4 (2.4–4.9) 2.1 (1.5–3.0) 2.1 (1.4–3.0)
Middle 1.6 (1.2–2.2) 1.1 (0.8–1.6) 1.1 (0.8–1.6) 2.3 (1.6–3.2) 1.7 (1.2–2.3) 1.6 (1.2–2.2)
High 1.0 1.0 1.0 1.0 1.0 1.0
Unknown 1.0 (0.7–1.3) 0.8 (0.6–1.1) 0.8 (0.6–1.1) 2.0 (1.4–2.9) 1.5 (1.1–2.3) 1.5 (1.0–2.1)
Education level
Less than secondary school 1.6 (1.2–2.2) 1.1 (0.7–1.6) 1.0 (0.7–1.6) 2.8 (2.1–3.7) 1.8 (1.3–2.4) 1.7 (1.3–2.3)
Some secondary school 1.1 (0.8–1.6) 1.0 (0.7–1.4) 1.0 (0.7–1.4) 1.4 (0.9–2.0) 1.2 (0.8–1.8) 1.2 (0.8–1.7)
Post-secondary school 1.0 1.0 1.0 1.0 1.0 1.0

We thank Ms Colette Koeune of Statistics Canada for her assistance in facilitating remote access to the NPHS longitudinal data. Dr Yue Chen currently holds a Canadian Institutes of Health Research Investigator Award. Dr Daniel Krewski is the NSERC/SSHRC/McLaughlin Chair in Population Health Risk Assessment at the University of Ottawa.

## References

1
King H, Aubert RE, Herman WH. Global burden of diabetes, 1995–2050: Prevalence, numerical estimates, and projections.
Diabetes Care

1998
;
21
:
1414
–24.
2
Wigle DT, Mao Y, Semenciw R, McCann C, Davies JW. Premature deaths in Canada: impact, trends and opportunities for prevention.
Can J Public Health

1990
;
81
:
376
–81.
3
Ali O, Tan TT, Asakinah O, Khalid BA, Wu LL, Ng ML. Prevalence of NIDDM and impaired glucose tolerance in aborigines and Malays in Malaysia and their relationship to sociodemographic, health, and nutritional factors.
Diabetes Care

1993
;
16
:
68
–75.
4
Sayeed MA, Ali L, Hussain MZ, Rumi MA, Banu A, Khan AKA. Effect of socioeconomic risk factors on the difference in prevalence of diabetes between rural and urban populations in Bangladesh.
Diabetes Care

1997
;
20
:
551
–55.
5
Hjelm K, Isacsson A, Apelqvist J, Sundquist J, Nyberg P. Foreign and Swedish-born diabetic patients—a population-based study of prevalence, glycaemic control and social position.
Scand J Soc Med

1996
;
24
:
234
–52.
6
Connolly V, Unwin N, Sherriff P, Bilous R, Kelly W. Diabetes prevalence and socioeconomic status: a population based study showing increased prevalence of type 2 diabetes mellitus in deprived areas.
J Epidemiol Community health

2000
;
54
:
173
–77.
7
Drury TF, Danchik KM, Harris MI. Sociodemographic characteristics of adult diabetics. In: Harris MI, Hamman RF (eds). Diabetes in America. 2nd Edn V. VII. Bethesda: US Department of Health and Human Services (PHS) 1995, pp. 1–37.
8
Millar WJ, Wigle DT. Socioeconomic disparities in risk factors for cardiovascular disease.
Can Med Assoc J

1986
;
134
:
127
–32.
9
James R, Young TK, Mustard CA, Blanchard J. The health of Canadian with diabetes.
Health Rep

1997
;
9
:
47
–52.
10
Statistics Canada. National Population Health Survey 1996–97. Public Use Microdata Files. Ottawa, Ontario, Canada: Health Statistics Division, 1997.
11
Chen Y, Dales R, Tang M, Krewski D. Obesity may increase the incidence of asthma in women but not in men: longitudinal observations from the Canadian National Population Health Surveys.
Am J Epidemiol

2002
;
155
:
191
–97.
12
Tang M, Chen Y. Prevalence of diabetes in Canadian adults aged 40 years or older (letter to editor).
Diabetes Care

2000
;
23
:
1704
–05.
13
World Health Orgnization: WHO Expert Committee on Diabetes mellitus. Second Report. Geneva: World Health Organization, 1999.
14
The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus: Report of the Expert Committee on the Diagnosis and classification of Diabetes Mellitus.
Diabetes Care

1997
;
20
:
1183
–97.
15
Harris MI, Goldstein DE, Flegal KM et al. Prevalence of Diabetes, impaired fasting glucose, and impaired glucose tolerance in U.S. adults: The third National Health and Nutrition Examination Survey.
Diabetes Care

1998
;
21
:
518
–24.
16
Rockwood K, Awalt E, MacKnight C, Mc Dowell I. Incidence and outcomes of diabetes mellitus in elderly people: report from the Canadian Study of Health and Aging.
Can Med Assoc J

2000
;
162
:
769
–72.
17
Potvin L, Richard L, Edwards AC. Knowledge of cardivocascular disease risk factors among the Canadian population: relationships with indicators of socioeconomic status.
Can Med Assoc J

2000
;
162
:
S5
–11.
18
van der Meer JB, Mackenbach JP. The care and course of diabetes: difference according to level of education.
Health Policy

1999
;
46
:
127
–41.
19
Unwin N, Watson W, Harland J. Social class differences in the prevalence of glucose intolerance. (Abstract).
Diabet Med

1995
;
12
:
S31
.
20
Stern MP, Rosenthal M, Haffner SM. Sex difference in the effects of sociocultural status on diabetes and cardiovascular risk factors in Mexican Americans. The San Antonio Heart Study.
Am J Epidemiol

1984
;
120
:
834
–51.
21
Stunkard AJ, Sorensen TI. Obesity and socioeconomic status—a complex relation.
N Engl J Med

1993
;
329
:
1036
–37.
22
Hays LM, Clark DO. Correlates of physical activity in a sample of older adults with type 2 diabetes.
Diabetes Care

1999
;
22
:
706
–12.
23
Ford ES, Merritt RK, Heath GW et al. Physical activity behaviors in lower and higher socioeconomic status populations.
Am J Epidemiol

1991
;
133
:
1246
–56.
24
Hamman RF. Genetic and environmental determinants of non-insulin-dependent diabetes mellitus (NIDDM).
Diabetes Metab Rev

1993
;
8
:
287
–338.
25
Kinsey JD. Food and families’ socioeconomic status.
J Nutr

1994
;
124
:
1878S
–85S.
26
Pomerleau J, Pederson LL, Ostbye T, Speechley M, Speechley KN. Health behavious and socio-economic status in Ontario, Canada.
Eur J Epidemiol

1997
;
13
:
613
–22.
27
Gray-Donald K, O’Loughlin J, Richard L, Paradis G. Validation of a short telephone administered questionnaire to evaluate dietary interventions in low income communities in Montreal, Canada.
J Epidemiol Community Health

1997
;
51
:
326
–31.
28
Kraut A, Tate R, Walld R, Mustard C. Impact of diabetes on employment and income in Manitoba, Canada.
Diabetes Care

2001
;
24
:
64
–68.
29
Chen Y, Rennie DC, Lockinger LA, Dosman JA. Association between obesity and high blood pressure: Reporting bias related to gender and age.
Int J Obes

1998
;
22
:
771
–77.