Diabetes Prevalence, Behavioural and Anthropometric Risk Factors, and Psychosocial Constructs in Three Aboriginal Communities in Central British Columbia
Mark Daniel, Diane Gamble, Joyce Henderson and Sandy Burgess

Abstract
To counter non-insulin-dependent diabetes mellitus (NIDDM) among aboriginal populations in the Okanagan region of British Columbia, a diabetes prevention and control project was initiated in 1994. This report presents the results of baseline diagnostic and risk factor screening among high-risk samples in three communities (n = 189). Persons with established diabetes and those at familial risk for NIDDM were measured for physiological, anthropometric and behavioural risk factors, as well as psychosocial constructs. Persons at risk underwent oral glucose tolerance tests. For adults 18 years and older, crude regional prevalence rates for diabetes and impaired glucose tolerance were 36.1 per 1000 and 7.8 per 1000, respectively. Proportions of individuals above the 70th percentile values from the Canada Fitness Survey for body mass index and waist-to-hip ratio were 60% and 73%, respectively. Of all persons sampled, cases of hypercholesterolemia (50%), hypertriglyceridemia (38%) and hyperinsulinemia (12%) were highly prevalent. Individuals with diabetes were characterized by poor control.

Key words: Aboriginal health; British Columbia; community screening; diabetes mellitus, non-insulin-dependent; prevalence; risk factors

Introduction

Recognizing the magnitude of the diabetes problem among native Canadian populations 6 and the corresponding need to collaborate with native people in developing, implementing and evaluating new models of health care delivery relevant to community needs,7 a community-based diabetes prevention and control project was initiated in 1994, in the central Okanagan region of British Columbia. The two-year project aims at developing a model to prevent and control diabetes, focusing on changing behaviours in combination with changing environments.8

From the platform of a population approach to prevention, a high-risk screening initiative was undertaken in three communities to identify persons with undiagnosed diabetes and to provide baseline data for evaluating the effectiveness of an intervention program developed and implemented in one of the three communities. Two follow-up tests are planned for each community. This report presents the results of baseline diagnostic assessments and physiological, anthropometric, psychosocial and behavioural risk factor screening activities done prior to implementing prevention and control strategies.

Methods

There are three points of measurement in each community, at evenly spaced intervals throughout the project. Based on the quasi-experimental non-equivalent control group design,9 changes over time in the intervention community will be evaluated against changes in the reference communities. All three communities can be classified in terms of language as Interior Salishan and in terms of culture as of the Plateau area.10

The Okanagan region falls into the South Mainland Zone quadrant as defined by the Medical Services Branch (MSB), Pacific Region, of Health Canada. Preliminary data gathered in 1987 showed the prevalence of diabetes among natives aged 35 and older in the South Mainland Zone (5.2%) to be greater than the provincial prevalence (4.5%) for the same age group.11 Health care is provided by MSB nurses with the support of community health representatives. Access to physician services is available in the towns of Vernon, Enderby and Penticton. Standard medical and nursing follow-up is provided in all communities for persons with established diabetes and for persons diagnosed with diabetes over the course of the project; the Okanagan intervention community receives the further benefit of prevention and control programs in the context of the project.

Participant Selection and Screening Protocol
Baseline tests in each community apply to high-risk samples comprising persons with established diabetes and first- and second-degree relatives of individuals with diabetes. Persons with established (physician-diagnosed) diabetes were identified using records maintained by the local MSB unit. Community workers hired on behalf of the project constructed family pedigrees for all individuals with established diabetes; they also identified persons related to non-residents with diabetes.

Persons with diabetes and those at familial risk for diabetes were asked to participate in a series of tests and measurements at the outset, midpoint and completion of the 24-month project. All participants provided their informed, written consent. Pregnant women and minors under age 18 were excluded from testing.

The research protocol was reviewed and approved at the University of British Columbia by two separate committees for research on human subjects: the Clinical Screening Committee and the Behavioural Sciences Screening Committee. The Okanagan University College Ethics Committee also approved the protocol.

All testing was conducted in community band halls between 7:30 am and 12:00 am. Testing was scheduled in each community for two sequential days, usually Friday and Saturday. The testing protocol followed this order: blood samples, anthropometric measurements, then administration of behavioural and psychosocial questionnaires.

Metabolic Variables
For screening at their band hall, participants provided a fasting venous blood sample (12-hour fast) and had their blood pressure taken using a calibrated aneroid sphygmomanometer. From all fasting blood samples, concentrations were determined for plasma glucose, insulin, glycosylated hemoglobin (HbA1c), plasma triglycerides, total cholesterol and high-density lipoprotein cholesterol (HDL).

For diagnostic purposes, a two-hour oral glucose tolerance test (OGTT) was administered to persons at risk for diabetes but not to those with established diabetes. Persons at risk were given a 75-gram carbohydrate load after their fasting blood samples were drawn, and gave another blood sample two hours later to assess post-load glucose levels. World Health Organization criteria 12 were applied to classify individuals on the basis of their OGTT as either normoglycemic (fasting and two-hour plasma glucose £ 7.7 mmol/L); having impaired glucose tolerance (IGT) [fasting plasma glucose £ 7.7 mmol/L and two-hour plasma glucose 7.8-11.0 mmol/L]; or having diabetes (fasting plasma glucose ³ 7.8 mmol/L and/or two-hour plasma glucose ³ 11.1 mmol/L). To estimate the prevalence of diabetes in each community, persons with previously established diabetes were pooled with persons diagnosed with diabetes on the basis of the screening procedure.

Anthropometric Variables
All measures were performed by registered nurses trained in anthropometric measurement. Participants wore light clothing (sweat pants and T-shirt), with footwear removed. Weight, height, five skinfolds, and waist and hip girths were assessed, taking the median of three measures as the "true" value.13 Clothing was adjusted to enable direct measurement of skinfolds and girths. Body mass index (BMI), the sum of five skinfolds (biceps, triceps, subscapular, iliac crest and medial calf), waist-to-hip girth ratio (WHR) and the sum of two trunk skinfolds (subscapular and iliac crest) were calculated.14 A computer program 15 was used to determine percentile ratings for these calculated measurements, in relation to age- and sex-dependent norms from the Canada Fitness Survey (1981).16

Behavioural Variables Physical activity was assessed on the basis of these questions: (a) "At least once a week, did you engage in any regular activity similar to brisk walking, jogging, bicycling, etc., long enough to work up a sweat?"; (b) "If yes, how many times per week?"; and (c) "What activity is this?"17 These questions have been validated as a measure of physical activity.18,19 Individuals were also classified according to graded categories of the frequency of sweat-producing activities per week. This approach is in keeping with epidemiologic studies suggesting a linear dose-response relationship between physical activity and health and functional effects, for low to moderate levels of activity.20

Dietary behaviour was assessed using records of food and drink consumed over three consecutive days, including one weekend day. Procedures for completing the food records were explained prior to testing. Charts and models were used to help estimate food portion sizes. Data were also collected on smoking and alcohol consumption.

Psychosocial Variables
Psychosocial instruments were chosen on the basis of brevity, simple language, acceptance and extent of use in divergent populations, and established reliability and validity.21 However, none of the instruments chosen had been validated among aboriginal populations, and a literature review suggested that neither the chosen measures nor any other psychosocial measures have been applied in population-based (i.e. other than clinical) settings, aboriginal or otherwise, to diabetes prevention and control.

Self-esteem was measured using Rosenberg's 10-question Guttman-type scale, which collapses into six items.22 Reproducibility, scalability and concurrent validity are well established.22 Scores correlate with clinical ratings and similar measures of self-esteem. Individuals rate their level of agreement with statements such as, "I feel I have a number of good qualities." Positive ratings were summed across items (maximum score = 6).

We assessed psychological well-being with Bradburn's 10-item Affect Balance Scale,23 for which reliability 23 and concurrent validity have been proven.24 Individuals answer "yes" or "no" to five positive and five negative statements. An example is, "During the last few weeks, did you ever feel particularly excited or interested in something?" Affirmative responses were counted, and the sum of responses for negative items subtracted from the sum for positive items. To correct for negative scores, a constant of 5 was added to the difference (maximum score = 10).

Mastery, the extent that people see themselves as being in control of the forces that affect their lives, was appraised using a 7-item, 5-point Likert-type mastery scale 25 with confirmed reliability and construct validity.26 Individuals rate their level of agreement on a series of seven positive and negative statements such as, "I can do just about anything I really set my mind to." Scores were averaged (maximum score = 5).

To assess social support we asked two questions that focused on emotional support: (a) "Among your friends and relatives, excluding your partner, if you have one, how many people do you feel you can tell just about anything to, people you can count on for understanding and advice?"; and (b) "If you live with a partner, is your partner someone you can really talk with about things that are important?"26 The first question was scored from 0 to 2, for replies of "no one," "one person," or "two or more." The second question was scored either 0 for a negative response or 1 for a positive response. Total scores thus ranged from 0 to 3.

Statistical Analysis
Summary statistics were computed for continuous (mean ± standard deviation) and categorical variables (number and percent). Community differences were tested by chi-squared test, Kruskal-Wallis test or analysis of variance (ANOVA), as appropriate. Analyses of relationships between physical activity, smoking status and fasting glucose and insulin levels were undertaken by ANOVA, using pooled data blocked on community, with a first order community by main effect interaction term included. Tests for differences in psychosocial variables between groups defined by behavioural characteristics were performed by ANOVA, controlling for age and sex. Stepwise multiple regression was used to determine which of the physiological, anthropometric and behavioural variables were significant predictors of glucose and insulin status. Analyses were conducted using the Statistical Package for the Social Sciences (SPSS).27

Results

In terms of basic characteristics, the high-risk samples in the three communities were similar (Table 1). In the Okanagan and Spallumcheen communities, 77% of persons sampled had enough income to buy nutritious foods, and 71% had enough income to engage in recreational physical activities. Affirmative responses for these questions were close to 50% for the Penticton sample.

Diagnostic procedures identified six newly diagnosed persons with diabetes among the Spallumcheen and Penticton samples, but none in the Okanagan sample (Table 2). However, six cases of IGT were identified in the Okanagan sample; four in the Spallumcheen and Penticton samples.The crude prevalence of diabetes (on-reserve population) was 2.8% for the Okanagan Band, 5.7% for the Spallumcheen Band and 2.9% for the Penticton Band. Across the three samples, approximately 75% of individuals were classified following screening as normoglycemic, 5% were diagnosed with IGT and 20% were classified as having diabetes. Sex did not vary with screening status: women averaged 60% across all categories; men, 40%.

The prevalence of high blood pressure (systolic blood pressure ³ 160 mmHg and/or diastolic blood pressure ³ 90 mmHg, and/or current treatment for established hypertension) was 10.4% across samples (Table 3). Percentage HbA1c was elevated among 27.3% of the Okanagan sample, whereas the elevated proportion was half of that (approximately 15%) among the Spallumcheen and Penticton samples. Hyperinsulinemia was observed among 12% of individuals across samples.

Across communities, almost 50% of persons sampled had high cholesterol levels (Table 3). High plasma triglyceride levels were observed among 30-50% of persons sampled, according to the community. Forward stepwise multiple regression analysis revealed that, of triglyceride, cholesterol and HDL concentrations, triglyceride was the sole predictor of fasting plasma glucose (F = 25.6, p < 0.001) and insulin (F = 5.7, p < 0.05) concentrations.

In terms of lifestyle and behaviour (Table 4), 65% of people across samples engaged in physical activity sufficient to work up a sweat at least once per week. These people had a lower mean level of insulin relative to those who did not take part in at least one sweat-producing activity per week (F = 7.7, p = 0.006). Only 37% of people sampled engaged in sweat-producing activities at least three times per week. Classifications of weekly sweat-producing activities were not related to fasting levels of glucose, HbA1c, cholesterol, triglycerides or HDL.

There were significant differences across samples in alcohol consumption (chi squared = 15.0 with 2 degrees of freedom [df], p = 0.02). Between 30% and 60% of people reported that they consumed alcohol and the median number of drinks they consumed weekly ranged from four to eleven (Table 4). Smoking status also varied significantly across samples (chi squared = 8.0 with 2 df, p < 0.001): 36-67% of individuals sampled were regular cigarette smokers. The average number of cigarettes smoked per day ranged from 11 to 16. Controlling for community, we did not observe significant relationships for either alcohol consumption or smoking status with fasting levels of glucose, insulin, cholesterol, triglycerides or HDL.

Three-day dietary food records revealed a high proportion of ambiguous and incomplete items in all community samples. Home visits and interviews are currently under way to clarify a large number of records. Food records will be analyzed to estimate the masses of protein, dietary fibre, carbohydrates and fat consumed, as well as total kilocalories and the percentage of caloric intake from protein, carbohydrates and fats. We will focus on general dietary behaviour as opposed to detailed nutritional status.

For anthropometric variables (Table 5), there were significant differences across communities for all but height and waist-to-hip girth ratio. Mean values for BMI ranged from 26.7 kg/m 2 to 30.6 kg/m 2 across communities. Mean values for WHR ranged from 0.90 to 0.92. For the sum of skinfolds (SOS), the range of mean values was 82-126 mm; for the sum of trunk skinfolds (SOTS), 44-68 mm. Based on norms 14 from the Canada Fitness Survey,16 the percentage of individuals above the 70th percentile for their age and sex was 59.2% for BMI, 60.1% for SOS, 72.8% for WHR and 66.8% for SOTS.

Forward stepwise multiple regression analysis with the variables BMI, WHR, SOS and SOTS showed that SOTS was the most powerful single predictor of insulin concentration, explaining 12.4% of the variation in insulin. After SOTS, WHR accounted for a further 3.0% of insulin variation. No further significant gain in variation of insulin could be achieved by incorporating BMI or SOS. Thus, the final stepwise regression equation involved two variables, SOTS and WHR, explaining 15.4% of the difference in insulin concentration (F = 16.0 with 2,175 df, p < 0.001). Analyzing the same explanatory variables as predictors of fasting glucose level showed that only WHR explained a significant portion of the variation (12.8%) in glucose concentration (F = 25.9 with 1,176 df, p < 0.001).

Scores were high in all samples for the psychosocial variables self-esteem, affect balance, mastery and social support (Table 6). Differences of marginal significance (p = 0.05) among samples appeared in affect balance and social support, while differences in self-esteem and mastery were insignificant. Fasting glucose level (but not insulin concentration) was predicted by mastery, which was inversely related to glucose levels (F = 6.4 with 1,158 df, p < 0.05). Self-esteem, affect balance and social support were not significant predictors of glucose or insulin levels. There were significant differences in scores for mastery (but not in those for self-esteem, affect balance or social support) between diabetes, IGT and normoglycemic groups (F = 5.4 with 2,170 df, p = 0.005).

Controlling for age and sex, scores for mastery (but not those for self-esteem, affect balance or social support) were significantly greater among non-smokers in contrast to current smokers (F = 8.4 with 1,175 df, p = 0.004); mean scores (±SD) were 3.68 (0.69) for current smokers and 3.87 (0.70) for non-smokers. Scores for psychosocial variables did not vary significantly across groups dichotomized in terms of alcohol consumption or sweat-producing activity, nor did psychosocial scores differ between groups defined by lipid status.

Discussion

In British Columbia,11 as in Saskatchewan 29 and Canada in general,30 relative urbanization is positively associated with the prevalence of diabetes among aboriginal peoples. Although increasing urbanization and acculturation have not been proven to cause diabetes, these changes, in combination with a genetic susceptibility for the development of NIDDM, are implicated in the rising prevalence of diabetes among First Nations peoples. Urbanization may explain the low rates of physical activity observed in this study; despite semi-rural locations and sustenance-based economies, only 37% of persons sampled engaged in sweat-producing activity at least three times per week. Physical activity three times per week is the threshold level recommended for beneficial effects on health.31 Our observations support targeting increases in low to moderate levels of physical activity, although 37% is not discordant with estimates of 40% for North Americans who engage in regular, moderate level physical activity.32

Our observation that persons who engaged at least once weekly in sweat-producing activity had a lower mean insulin level than persons who did not (p < 0.01) is consistent with a large body of literature associating physical activity with greater insulin sensitivity in skeletal muscle and adipose tissue.3 That we did not observe a significant trend over graded categories of weekly exercise in effects on clinical variables most likely reflects the crudeness of our measure of physical activity.

Reporting a greater degree of habitual physical activity related to food gathering and agricultural practice in cultures maintaining a traditional lifestyle, Zimmet et al. postulated that the adoption of sedentary lifestyles is an integral aspect of the acculturation of indigenous peoples, from which arises hyperinsulinemia, insulin resistance and ultimately glucose intolerance, particularly in those who are obese.33 Participation in Zuni fitness promotion programs has been associated with reductions in weight and improvements in glycemic control among individuals both with and without diabetes.34 The relationship between diabetes and alcohol consumption is well documented,35 but levels of alcohol consumption were low across our three samples and do not suggest a need for action. We did not find significant relationships between alcohol status and any clinical variable. However, a limitation of our methodology was the use of self-reports to ascertain the prevalence of behavioural risk factors. The validity of indications of alcohol consumption is unclear, and self-reports of the number of drinks per week may be biased in that 25% of persons indicating alcohol consumption did not respond to a question about the average number of drinks per week.

In contrast, cigarette smoking was sufficiently prevalent to support intervention activities. The absence of relationships between smoking status and clinical variables may reflect the fact that historic data were not gathered on cigarette smoking; therefore, we were not able to estimate and analyze pack-years. The prevalence of high blood pressure in the Okanagan sample supports a need for intervention since hypertension has been shown to increase the risk of NIDDM independently of other known risk factors, including obesity.19

Poor control of diabetes across communities is indicated by the finding that 63% of all cases of elevated HbA1c were observed among people with diabetes (24 of 38 total cases-not shown in Table 3). Dietary behaviour among persons with diabetes appears oriented toward the consumption of foods overly high in fat and simple carbohydrates: persons with diabetes accounted for 35% of all cases of hypertriglyceridemia, 20% of all cases of hypercholesterolemia and 63% of all cases of elevated HbA1c.

Beyond targeting improved control for diabetics, the need for prevention is apparent for persons with IGT and normoglycemic profiles, who together accounted for 67% of all cases of hyperinsulinemia, 80% of all cases of hypercholesterolemia and 65% of all cases of high triglyceride levels. Normoglycemic individuals and persons with IGT also seem to have high levels of fat consumption. In the Okanagan sample, a high proportion of normoglycemic individuals with clinically elevated HbA1c (16%) suggests excessive consumption of simple carbohydrates in addition to fats. Though the contribution of dietary fats or a shift from traditional native foods to a Western high-fat diet is not clear in the development of diabetes,1 dietary intervention is supported by our finding that triglyceride concentration predicted fasting plasma glucose (p < 0.001) and insulin levels (p < 0.05).

Obesity is considered the most powerful risk factor for NIDDM,2 and BMI is a predictor of diabetic status, fasting plasma glucose and glycosylated hemoglobin in native Canadian populations.36 In this study, mean values for BMI and SOS were elevated in all samples beyond Canadian norms and cut-offs for health risk.14 Nevertheless, despite the fact that BMI and SOS each explained significant portions of the variation in glucose and insulin levels when examined apart from WHR and SOTS, the latter variables (which describe abdominal and trunkal adiposity, respectively) consistently eclipsed variation explained initially by BMI or SOS in multivariate analyses, to the extent that BMI and SOS could not account for any significant portions of variation in glucose and insulin levels beyond that explained by WHR or SOTS. Therefore, we conclude that for the persons with diabetes and the high-risk individuals sampled in this study, an abdominal or trunkal predominance of adipose tissue is better related to glucose and insulin levels than indicators of overall obesity.

Regarding body fat distribution as a risk factor for NIDDM, an abdominal or trunkal pattern has been established prospectively as a diabetes risk factor greater than and independent of obesity.37 Among native Canadian populations, trunkal adiposity has been associated with elevated glycosylated hemoglobin levels,38 glucose intolerance and diabetes.39 In this study, whereas 59% of high-risk individuals sampled were above the 70th percentile for BMI for their age and sex, 73% were above the 70th percentile for WHR for their age and sex.

In keeping with the notion that psychosocial factors describing the way people feel are important in relation to diabetes prevention and control, our observation that fasting glucose level was predicted by mastery (p < 0.05) and that this relationship is inverse in nature has meaningful implications for community-based intervention programs. The positive linear trend in mastery across groups defined in terms of clinical diabetic status (p = 0.005) has further implications for community-based intervention, though neither relationship can be construed as causal. We do not know whether low mastery scores are a function of physiology or clinical status or if high glucose levels and predisposing behaviours are a function of a low sense of mastery.

Nevertheless, as an indication of the degree to which individuals feel in control of the forces that affect their lives, mastery is in part a function of social, economic and political realities. Targeting changes in social and cultural factors that bear on mastery might increase the prevalence of health promoting behaviours, as well as adherence to such behaviours. This logic and the preceding observations thus support a focus on community-based, culturally appropriate methods of preventing and controlling diabetes.

The literature contains numerous testimonies to the limited effectiveness of orthodox health care approaches as applied to address diabetes in indigenous populations. Community-based, culturally sensitive diabetes interventions are likely to be associated with positive changes in the psychosocial impact of the disease. In this way psychosocial constructs are relevant to community-based interventions not as clinical indications of status nor as a basis for inferential comparisons with other populations, but as a means of monitoring change over time in individual and collective capacities that constitute the change mechanism by which programs might prevent and control diabetes. This project is founded upon the notion that a program developed, implemented and controlled by community members will aid and promote personal and collective efficacy, whereby people work together to effect change at the community level. The theoretical basis of this approach is Social Learning Theory.40

Recognizing that a program theory is also an evaluation model, we selected methods and measures to correspond to components of the program theory. The evaluation hypothesis is that a community-based, culturally appropriate diabetes prevention and control program will invoke positive psychosocial responses that correlate over time with positive behavioural changes, which correlate in turn with positive physiological and anthropometric changes. In this project, psychosocial variables relate to the concept of self-efficacy.40 Changes in mastery, affect balance, self-esteem and social support, in relation to changes in behaviour and physiological variables, could provide compelling support for interventions based on contextual issues identified by community members.

For the communities involved in this project, the prevalence of diabetes and IGT and the prevalence of risk factors among the individuals sampled support the need for action in the form of prevention and improved control. The fact that community-based research involves multiple methodological challenges, and that such undertakings can be critiqued on the basis of confounded variables or lack of reliability, ought not to preclude the need to develop, implement and evaluate, as rigorously as possible, integrated intervention strategies to address the growing problem of diabetes in First Nations communities.

Positive behavioural and environmental change is a matter of capacity-building and channelling strengths into appropriate actions, and there is no better catalyst than an empowered community. Empowerment is not something that can be done to people; people become empowered when they are able to help themselves. Community-based, participatory research can aid the process of empowerment among individuals and groups.7 In this way, personal and collective efficacy can be targeted and used to good advantage. Diabetes prevention and control may be possible through the use of strategies that are acceptable, meaningful and relevant to those whose lives are closest to the disease.

Acknowledgements

The authors acknowledge the contributions of the following members of the research team: Beatrice Bonneau, Laura Miller and Sadie Muik, Okanagan Indian Band, Vernon, BC; Linda Mitchell, Spallumcheen Indian Band, Enderby, BC; and Bernice Strachan, RN, Diabetes Day Programme, Vernon Jubilee Hospital, Vernon, BC. The assistance and co-operation of community participants and Advisory Committee members, and the Okanagan, Penticton and Spallumcheen Band Councils, is acknowledged with appreciation. The authors extend special thanks for technical assistance to Lowell Laidlaw and Betty Carlson, Clinical Chemistry Laboratory, Vernon Jubilee Hospital, and to student volunteers from the BSN program at Okanagan University College.

References

2. World Health Organization. Expert Committee on Diabetes Mellitus. Second report. Geneva: World Health Organization, 1980; Technical Report Series No 646.

3. Ruderman N, Apelian AZ, Schneider SH. Exercise in therapy and prevention of type II diabetes. Implications for blacks. Diabetes Care 1990;13(11):1163-8.

4. Stern MP. Kelly West Lecture. Primary prevention of type II diabetes mellitus. Diabetes Care 1991;14(5):399-410.

5. Green LW, Kreuter MW. Health promotion planning: an educational and environmental approach. Mountain View (CA): Mayfield, 1991.

6. Daniel M, Gamble D. Diabetes and Canada's aboriginal peoples: the need for primary prevention. Int J Nurs Stud 1995;32(3):243-59.

7. Green LW, George MA, Daniel M, Frankish CJ, Herbert CJ, Bowie WR, et al. Participatory research in health promotion: review and recommendations for the development of participatory research in health promotion in Canada. Ottawa: The Royal Society of Canada, 1995; ISBN 0-920064-55-8.

8. Daniel M, Green LW. Application of the precede-proceed planning model in diabetes prevention and control: a case illustration from a Canadian aboriginal community. Diabet Spectrum 1995;8(2):74-84.

9. Campbell DT, Stanley JC. Experimental and quasi-experimental designs for research on teaching. In: Gage NL, ed. Handbook of research on teaching. Chicago: Rand McNally and Company, 1963.

10. Ministry of Native Affairs. The aboriginal peoples of British Columbia: a profile. Victoria: Province of British Columbia, 1990; C90-092086-6.

11. Martin JD, Bell P. Diabetes mellitus in the native population of British Columbia, Canada. Vancouver: Medical Services Branch (Pacific Region), Health and Welfare Canada, 1990.

12. World Health Organization. Diabetes mellitus: report of a WHO study group on diabetes mellitus. Geneva: World Health Organization, 1985; WHO Technical Report Series No 727.

13. Ross WD, Marfell-Jones MJ. Kinanthropometry. In: McDougall SD, Wenger HA, Green HA, McDougall SD, Wenger HA, Green HA, eds. Physiological testing of the elite Athlete. Ottawa: Mutual, 1982:75-115.

14. Canadian Standardized Test of Fitness. CSTF operations manual. 3rd ed. Ottawa: Canadian Association of Sport Sciences, 1983.

15. Daniel M. Percentiles for Canadian Standardized Test of Fitness norms and associated health risk zones for anthropometric indicators of relative adiposity and adipose distribution [computer program, Version 1.00 for IBM-DOS]. Burnaby (BC): Faculty of Applied Sciences, Simon Fraser University, 1987.

16. Health and Welfare Canada. Canada Fitness Survey. Ottawa: Health and Welfare Canada, 1981.

17. Manson JE, Rimm EB, Stampfer MJ, Colditz GA, Willett WC, Krolewski AS, et al. Physical activity and incidence of non-insulin-dependent diabetes mellitus in women. Lancet 1991;338(8770):774-8.

18. LaPorte RD, Black-Sandler R, Cauley J. The assessment of physical activity in older women: analysis of the interrelationship and reliability of activity monitoring, activity surveys, and caloric intake. J Gerontol 1983;38:394-7.

19. Helmrich SP, Ragland DR, Leung RW, Parffenbarger RJ. Physical activity and reduced occurrence of non-insulin-dependent diabetes mellitus. N Engl J Med 1991; 325(3):147-52.

20. Blair SN, Kohl HW, Gordon NF, Paffenbarger RSJ. How much physical activity is good for health? Ann Rev Pub Health 1992;13(99):99-126.

21. Noack H. Conceptualising and measuring health. In: Badura B, Kickbusch I. Health promotion research: towards a new social epidemiology. Copenhagen: World Health Organization Regional Publications, 1991:85-112; European Series No 37.

22. Rosenberg M. Society and the adolescent self-image. Princeton: Princeton University Press, 1965.

23. Bradburn NM. The structure of psychological well-being. Chicago: Aldine, 1969.

24. Moriwaki SY. The affect balance scale: validity study with aged samples. J Gerontol 1974;29:73-8.

25. Pearlin LI, Schooler C. The structure of coping. J Health Soc Behav 1978;19:2-21.

26. Pearlin LI, Lieberman MA, Menaghan EG, Mullan JT. The stress process. J Health Soc Behav 1981;22:337-56.

27. Statistical package for the social sciences SPSS. Statistical data analysis, Version 4.0 [computer program]. Chicago: SPSS Incorporated, 1990.

28. Fox C, Harris SB, Whalen-Brough E. Diabetes among native Canadians in northwestern Ontario: 10 years later. Chronic Dis Can 1994;15(3):92-6.

29. Dyck RF, Tan L, Hoeppner VH. Body mass index, gestational diabetes and diabetes mellitus in three northern Saskatchewan aboriginal communities [short report]. Chronic Dis Can 1995;16(1):24-6.

30. Young TK, Szathmary EJE, Evers S, Wheatley B. Geographical distribution of diabetes among the native population of Canada: a national survey. Soc Sci Med 1990;31(2):129-39.

31. American College of Sports Medicine. Position statement on the recommended quantity and quality of exercise for developing and maintaining fitness in healthy adults. Med Sci Sports Ex 1990;22:265-74.

32. Stephens T, Jacobs DR, White CC. A descriptive epidemiology of leisure-time physical activity. Public Health Rep 1985;100:147-58.

33. Zimmet P, Dowse G, Finch C, Serjeantson S, King H. The epidemiology and natural history of NIDDM-lessons from the South Pacific. Diabet Metab Rev 1990;6(2):91-124.

34. Leonard B, Leonard C, Wilson R. Zuni diabetes project. Public Health Rep 1986;101(3):282-8.

35. Mohs ME, Leonard TK, Watson RR. Interrelationships among alcohol abuse, obesity, and type II diabetes mellitus: focus on native Americans. World Rev Nutrit Dietet 1988;56(93):93-172.

36. Young TK, Sevenhuysen G, Ling N, Moffatt M. Determinants of plasma glucose levels and diabetic status in a northern Canadian Indian population. Can Med Assoc J 1990;142:821-30.

37. Bergstrom RW, Newell-Morris LL, Leonetti DL, Shuman WP, Wahl PW, Fujimoto WY. Association of elevated fasting C-peptide level and increased intra-abdominal fat distribution with the development of NIDDM in Japanese-American men. Diabetes 1990;36:179-86.

38. Young TK, Sevenhuysen G. Obesity among northern Canadian Indians: patterns, determinants and consequences. Am J Clin Nutr 1989;49:786-93.

39. Szathmary EJE, Holt N. Hyperglycemia in Dogrib Indians of the Northwest Territories, Canada: association with age and a centripetal distribution of body fat. Hum Biol 1983;55(2):493-515.

40. Bandura A. Social Learning Theory. Englewood Cliffs: Prentice-Hall, 1977.

Author References

This paper presents further information from a project that was presented at the 3rd International Conference on Diabetes and Indigenous Peoples: "Theory, Reality and Hope," held in Winnipeg, Manitoba, May 26-30, 1995.