Predicting Absolute Risk of Type 2 Diabetes Using Age and Waist Circumference Values in an Aboriginal Australian Community
Predicting Absolute Risk of Type 2 Diabetes Using Age and Waist Circumference Values in an Aboriginal Australian Community
Odewumi Adegbija 0 1
Wendy Hoy 0 1
Zhiqiang Wang 0 1
0 Centre for Chronic Disease, School of Medicine, University of Queensland , Brisbane , Australia
1 Academic Editor: Kumarasamy Thangaraj, Centre for Cellular and Molecular Biology, INDIA
Funding: This project was supported by National
Health and Medical Research Council of Australia
APP1025350). The baseline screening was
supported by National Health and Medical Research
Council of Australia grant (320860). Dr Wendy Hoys
National Health and Medical Research Council of
Australia Australia Research Fellowship (#511081).
The funders had no role in study design, data
collection and analysis, decision to publish, or
preparation of the manuscript.
To predict in an Australian Aboriginal community, the 10-year absolute risk of type 2
diabetes associated with waist circumference and age on baseline examination.
A sample of 803 diabetes-free adults (82.3% of the age-eligible population) from baseline
data of participants collected from 1992 to 1998 were followed-up for up to 20 years till
2012. The Cox-proportional hazard model was used to estimate the effects of waist
circumference and other risk factors, including age, smoking and alcohol consumption status, of
males and females on prediction of type 2 diabetes, identified through subsequent
hospitalisation data during the follow-up period. The Weibull regression model was used to calculate
the absolute risk estimates of type 2 diabetes with waist circumference and age
Of 803 participants, 110 were recorded as having developed type 2 diabetes, in subsequent
hospitalizations over a follow-up of 12633.4 person-years. Waist circumference was
strongly associated with subsequent diagnosis of type 2 diabetes with P<0.0001 for both genders
and remained statistically significant after adjusting for confounding factors. Hazard ratios
of type 2 diabetes associated with 1 standard deviation increase in waist circumference
were 1.7 (95%CI 1.3 to 2.2) for males and 2.1 (95%CI 1.7 to 2.6) for females. At 45 years of
age with baseline waist circumference of 100 cm, a male had an absolute diabetic risk of
10.9%, while a female had a 14.3% risk of the disease.
The constructed model predicts the 10-year absolute diabetes risk in an Aboriginal
Australian community. It is simple and easily understood and will help identify individuals at risk of
Competing Interests: The authors have declared
that no competing interests exist.
diabetes in relation to waist circumference values. Our findings on the relationship between
waist circumference and diabetes on gender will be useful for clinical consultation, public
health education and establishing WC cut-off points for Aboriginal Australians.
Diabetes is one of the fastest growing chronic conditions in Australia, with an estimated 280
people developing the disease daily . Indigenous Australians (Aboriginals and Torres Strait
Islanders) [2, 3] are at higher risk of developing the disease, and at earlier ages  than the
non-indigenous group. A large proportion of Aboriginals develop type 2 diabetes (T2D) in
their lifetime, with a lifetime risk in one community of one in two among men and two in three
among women . Despite the preventive strategies aimed at controlling the development of
type 2 diabetes through healthy diet and lifestyle or medication , there has been little, if any
reduction in the high prevalence, high burden of mortality and complications imposed by this
disease among Aboriginals .
Several studies have shown relationship between excess abdominal fat and increased risk of
T2D [1, 9, 10]. In Australia, Aboriginals have the propensity for excessive abdominal fat which
reflected in their higher waist circumference (WC) compared to non-Aboriginals [11, 12].
Although, studies conducted in some Aboriginal communities have shown WC was a better
predictor of T2D compared to body mass index (BMI) and waist-to-hip ratio (WHR) [13, 14], but
WC cut-off points for Aboriginals to alert them of the risk of T2D and other chronic diseases
have not been established. Prediction models for the risk of diabetes have been developed in a
number of populations with the aim of predicting diabetes occurrences, while providing
intervention . Indeed, the Australian type 2 diabetes risk assessment tool (AUSDRISK)
included WC and ethnicity (Aboriginal or non-Aboriginal) as risk factors for estimating the risk
of developing diabetes, suggesting the differences in the effects of WC on diabetes between
Aboriginals and other Australians [18, 19]. However, due to low numbers of indigenous
Australians in the AUSDRISK study, Aboriginals were grouped with southern Europeans and Asians
to generate a high-risk group, which does not give a true representation of the level of diabetic
risk among Aboriginals in Australia. In this study, we reported the first 10-year absolute risk
estimates of diabetes using WC and age in a remote Australian Aboriginal community. As
there are no specific WC cut-off thresholds for Aboriginals in Australia, we categorised WC
into gender-specific quartiles for analysis and developed a model using WC and age values. To
quantify the impact of WC on the risk of developing diabetes, we developed a simplified tool
that can be used by health professional and the general public to understand how diabetes risk
varies with WC values. This tool can also be used to educate and alert individuals of the risk of
developing diabetes according to WC and age. Furthermore, this tool will also be helpful for
the planning and conducting obesity-related health education programs for the prevention and
management of T2D in Aboriginal communities in Australia.
The baseline characteristics of the study population have been described in detail elsewhere
[13, 20]. From January 1992 to December 1998, a total of 935 adults (> = 18 years, over 80%
recruitment) were included in a community-wide screening program in a remote Aboriginal
community in Australias Northern Territory. Written informed consents were obtained from
all participants at baseline measurements. The baseline database (containing screening of
anthropometric measurements) was merged with hospitalisation records to identify type 2
diabetes outcomes according to patients hospital registration numbers (HRN). Of the 935
individuals, 803 were free of T2D at baseline examination, and were followed up on hospital
records for up to 20 years from 1st February 1992 to 31st May 2012. Follow-up stopped for an
individual once he/she developed T2D or died. Prior to using the data, each participant was
deidentified and given a unique study ID number (SIN). This original baseline database was
approved by the Aboriginal community and Ethics Committee of the Menzies School of Health
Research and Territory Health Services. The project was approved by the Behavioural and
Social Science Ethical Review Committee of the University of Queensland (#2011001232).
Participants were followed-up through hospitalization records. Each participant was identified
by hospital registration number (HRN) and the patients health record ID codes. Our outcome
was newly diagnosed (incident) T2D as recorded in hospitalization data records. We identified
individuals with T2D using the International Classification of Diseases (9th revision; ICD-9)
code 250 and (10th revision, ICD-10) code E11 as recorded in the hospitalization dataset.
Follow-up period for participants with incident T2D was the time from the baseline survey date to
the diagnosis date. For individuals who did not develop T2D, the follow-up period was the
interval between the baseline screening and the follow-up time.
Waist circumference (WC) was measured in centimetres (cm) at baseline screening and
grouped into gender-specific quartiles for analysis. Quartiles for males: (Q1 = 6378 cm,
Q2 = 7985 cm, Q3 = 8695 cm, Q4 = 96138 cm). Quartiles for females: (Q1 = 6079 cm,
Q2 = 8090 cm, Q3 = 91101 cm, Q4 = 101.5135 cm). Q1 was the reference group for
comparison. For the T2D absolute risk prediction, we included WC and age in our model as they
have been identified as risk factors for diabetes in Aboriginals [13, 14, 21] and other
populations outside Australia [22, 23].
Continuous variables were expressed as the mean +/- SD as appropriate. Categorical data were
expressed as frequencies and percentages. To assess the association of baseline WC measures
on the newly diagnosed T2D, the Cox proportional hazards models were used to estimate
hazard ratios (HRs), adjusting for three confounding factorsage, smoking status and alcohol
consumption status. Age (years) was included in the analysis as a continuous variable, smoking
and alcohol status as categorical variables. The hazard ratios were computed for quartiles Q2,
Q3 and Q4 as compared with the lowest quartile (Q1) in different Coxs proportional hazards
regression models. To compare the associations of WC with T2D between males and females,
we converted original WC, BMI and WHR values into gender specific z scores for both genders
while also controlling for age, smoking and alcohol consumption status. The Weibull
regression model was used to predict an individuals 10-year T2D risk in adult males and females
using the formula:
where 0 represented the baseline WC coefficient, j was the coefficient for covariates (WC and
age), Xj represented the covariates, t = time and p = the shape parameter. We constructed the
regression coefficient based model by assigning values as estimated regression coefficients.
WC and age were fitted as continuous variables for the estimating the absolute risks of T2D.
For all analyses, two-tailed p values of <0.05 were considered significant. All statistical
analyses were performed with STATA version 12.0  and analysis were done separately for males
and females. S1 Dataset contains both primary and additional data used for analysis, while S1
Text shows STATA commands used in generating additional data variables used for analysis.
A total of 803 adults who were free from diabetes at baseline records were included in the
analysis. The baseline characteristics of participants are shown in Table 1. One hundred and ten
participants (38 males and 72 females) were diagnosed as having new onset diabetes during the
follow-up period of 12633.4 person-years. The median follow-up time was 18 years. The overall
diabetes incidence rate was 8.7 (95% CI: 7.210.5) per 1,000 person-years, 5.6 (95% CI: 4.17.7)
per 1,000 person-years for males and 12.2 (95% CI: 9.715.4) per 1,000 person-years for females.
Fig 1 shows the cumulative incidence of diabetes (%) according to WC quartiles for males and
females respectively in the follow-up time. As WC increased, the cumulative incidence also
increased to about 18% for males and 37.5% for females in WC Q4 during the follow-up period.
Hazard ratios from Coxs proportional hazards regression models
The crude HR for developing T2D with WC as a continuous variable was not significantly
different between males and females. Hazard ratio (HR) for 1 cm increase in WC was 1.04 (95%
CI: 1.021.06, p<0 0001) for males and 1.05 (95%CI: 1.041.07, p<0.0001) for females. Crude
and adjusted hazard ratios (HR) for WC quartiles are shown in Table 2. The observed median
follow-up times were 18.6, 18.2, 18.1 and 17.3 years for males and 18.2, 18.1, 16.8 and 15.9
years for females in WC quartiles Q1, Q2, Q3 and Q4 respectively.
The HR was 6.8 (95%CI: 2.023.4) for males in the highest (Q4) WC quartile compared to
those in the lowest (Q1) WC quartile, with p = 0.0002. After adjusting for confounding
variables of age, smoking and alcohol status, this association remained statistically significant
(HR = 5.9, 95%CI: 1.621.4, p = 0.0003). The crude HR for females in Q4 WC quartile was
Fig 1. Gender-specific cumulative incidence of diabetes(%) by waist circumference quartiles.
HR = 7.6, (95%CI: 3.218.0, p<0.0001). Again, after adjusting for confounding variables, the
association remained statistically significant for females, with HR = 7.2 (95%CI: 3.017.4,
Table 3 shows the crude and adjusted HR of T2D corresponding to 1 standard deviation
increase in WC, BMI and WHR (z-scores). The crude hazard ratio for males were 1.7 (95%CI:
1.32.2), 1.6 (95%CI: 1.22.1) and 1.3 (95%CI: 1.01.6) for WC, BMI and WHR respectively.
For females, corresponding WC, BMI and WHR crude hazard ratios were 2.1 (95%CI: 1.7
2.6), 1.9 (95%CI: 1.62.4) and 1.3 (95%CI: 1.01.6). Associations remained statistically
significant independently for WC and BMI in both genders after controlling for age, smoking and
alcohol consumption. However, the interaction term between gender and WC was not
statistically significant (p = 0.22). Likewise, incorporating an interaction term between age and
WC was not statistical significant (p = 0.08).
*Adjusted for age, smoking status and alcohol drinking status.
Waist circumference (WC). Quartile 1 (Q1). Quartile 2 (Q2). Quartile 3 (Q3). Quartile 4 (Q4).
HR (Hazard ratio); 95%CI (95% confidence interval
*Adjusted for age, smoking status and alcohol drinking status.
HR (Hazard ratio); 95%CI (95% confidence interval.
Absolute risk of diabetes by WC and age
Absolute risk Males
Absolute risk Females
Based on the coefficients of the final Weibull models above, we estimated 10-year absolute
risks according to age and WC values. This was illustrated in Fig 2, showing a 10-year absolute
diabetes risk (incidence of a first T2D event) of males and females respectively at different WC
and age values. Diabetes risk increased with higher WC and older age. For males, the lowest
diabetes risk (< = 2%) was for those younger than 30 years with WC less than 73 cm. For
females, those with WC less than 75 cm at any age presented the lowest diabetes risk (< = 2%).
Absolute diabetes risk of 3234% was observed in males over 60 years with WC greater than
110 cm and for females over 55 years of age with WC greater than 113 cm.
Using long-term cohort data from a remote community, we have developed a simple model to
estimate 10-year risk of T2D in an Aboriginal Australian community based on two variables:
WC and age. The results are based on a maximum of 20 years follow-up (median, 18 years)
and ascertainment of T2D incidence cases. The values for the two predictors used can easily be
obtained by individuals and in clinical practice. The absolute risk chart we developed is simple
to use and understand. The availability of the simple tool to predict future risk of T2D should
improve the understanding of WC on T2D risk and identify high-risk individuals based on
WC and age. It can also serve as an adjunct for planning and conducting public health
education programs for T2D and augment preventive strategies for Aboriginal communities.
Our approach based on the absolute risk method calculated from Weibull regression takes
into account the synergetic effects of risk factors of disease of interest . This approach took
into account the different follow-up time of all participants in the study to predict the risk of
the disease. We have presented the prediction to a 10 year period as the goal of this study is to
Fig 2. Gender-specific absolute risk(%) of type 2 diabetes, using waist circumference(cm) and age(years).
assess how risk of T2D change with WC values with a given age to assist in alerting individuals
of the risk of T2D with increasing WC. While T2D is a complex multifactorial disease, and the
cause originates from interactions among a number of risk genetic and environmental
predictors, increased risk in Australia has been associated with increasing age, family history of
diabetes, overweight (particularly with extra weight around the waist region), of indigenous
Australian descent, lack of physical activity, unhealthy diet and high blood pressure .
Unfortunately, we have no available data for family history of diabetes, physical activity and diet.
The focus of this study was to calculate gender-specific absolute risk estimates for a community
of Aboriginal Australians according to WC and age while adjusting for age, smoking and
drinking in the multivariate analysis. Central obesity measured by WC has been known to
increase the risk of T2D in the Australian Aboriginal population [6, 13]. In the present study, we
showed WC was a slightly better predictor of T2D than BMI and was much better at predicting
T2D than WHR. This is consistent with the study conducted by Wang and Hoy (2004) in the
same community presenting the odds ratio for diabetes after adjusting for age and sex to be
2.16, 1.80, 1.41, 1.81 and 1.84 for WC, BMI, weight, WHR, and hip circumference, respectively
. As there are no WC cut-off points for Australian Aboriginals, we categorised WC into
gender-specific quartiles for the analysis and presented a range for WC (65115 cm) and age
(2565 years) for the prediction to enable easy use of the information from this study. Age has
also been a commonly used single risk factor for detecting undiagnosed diabetes, more
importantly, when used in addition to one or more risk factors such as obesity, family history of
diabetes and hypertension, has the capacity to identify more individuals with undiagnosed
diabetes . An Australian study assessed the relationship of increasing prevalence of diabetes
with population ageing and obesity . They considered age group and BMI classification,
and found that the greatest relative percentage increases over time were observed among those
with normal BMI aged 60 years or older (148%), and those who were obese and less than 60
years of age (139%). While BMI is mostly widely used in Australia, very few studies give reports
of WC measured.
A previous Australian diabetes-prediction tool (Australian type 2 diabetes risk assessment
tool- AUSDRISK) identified WC, age and ethnicity as risk factors in the prediction of incident
diabetes [18, 19]. However, Aboriginals in the AUSDRISK were grouped with Southern
Europeans and Asians to generate a high-risk group which does not reflect a true representation of
the level of diabetic risk among Aboriginals in Australia. Therefore, the validity and
applicability of the tool to Aboriginals in Australia is questionable as they were derived from populations
with different risk-factor profiles and ethnicities. The uniqueness of our study lies on our focus
on one Aboriginal community; to estimate how the diabetes risk varies according to WC and
age values, both have been reported as predictors of diabetes [13, 29].
In our study, the lowest T2D incidence was observed among persons in the lowest WC
quartile, and we observed the strongest relationship between WC and T2D in the highest WC
quartile. There was statistical significant association between WC and diabetes for both males
and females in this population. Our estimates of cumulative incidence in Fig 1 suggest a high
incidence of diabetes with high WC particularly among females (37.5%). Based on point
estimates, our findings revealed females had higher WC values, and were at higher risk of T2D
with increasing WC compared to males. The interaction terms incorporated among the
variables used showed no statistical significance. A potential reason for this could be our relatively
small sample size resulting in insufficient power to our study in detecting the interactions.
Strengths and limitations
Our study has a number of strengths. First, to our knowledge, this is the first diabetes
prediction model specifically developed for an Aboriginal population in Australia. We focused on
Aboriginals in a remote Aboriginal community where they were culturally homogenous
(>80% ascertainment of the population). Also focusing on one Aboriginal community
minimised the impact of heterogeneity in body habitus, as epidemiological studies on WC in the
indigenous population of Australia have shown that there is substantial variation across
communities . Second, the prospective study design used minimised systematic error
introduced by the recall bias that cross-sectional and case-control studies are subject to. With a
follow-up period of up to 20 years with high participation and follow-up rates, there was a robust
ascertainment of diabetes events.
A few limitations of our study need to be acknowledged. First, T2D was only ascertained
through recording of a T2D diagnosis reported among the diagnoses for a hospitalization
episode. It is likely that some people with diabetes were not hospitalised in the community during
the study period, so the absolute risk presented in this study could underestimate the true risk
in the population. Second, data on other important risk factors of diabetes such as physical
activities, diet and family history of diabetes were not available. Therefore, we could only analyse
on data available to us (age, WC, smoking, gender, alcohol consumption) in our analysis.
Potential confounding effects of other variables (physical activities, family history of diabetes and
diet) were not adjusted in the reported strong association between WC and diabetes risk in this
study. Third, since our data were from one community, we were unable to generalise our
findings about WC-associated absolute risks of T2D in other Australian Aboriginal communities.
Due to our relatively small sample size (803 participants and 110 newly developed T2D), we
could not assess internal validation by using subset of individuals from the study sample.
However, our findings can be further replicated in other Aboriginal communities to assess the
generalizability of our findings. Fourth, there may have been inaccuracies of WC measurements at
baseline, resulting in the misclassification of participants into WC quartiles and attenuating
the observed associations for WC. Finally, the nature of survival analysis with different
followup time periods for different participants did not permit the calibration of our predicted risks.
However, we have shown the gender-specific cumulative incidence for the WC quartiles to
observe the proportion of individuals who had diabetes in the follow-up period according to
We have constructed a simple tool for predicting the 10- year diabetes risk using WC and age
as covariates, and this model focused on Aboriginals in an Australian remote community. This
simple tool would assess how absolute risk of T2D changes with WC values. It is also helpful
for identifying high-risk individuals, and developing strategies for preventing diabetes in
Aboriginal Australians. Our prediction tool will benefit from further validation with the inclusion
of other important risk factors such as family history of diabetes, physical activity and diet.
S1 Text. STATA commands used for generating secondary data in the dataset.
We thank the Aboriginal people who participated in this study. This project was supported by
National Health and Medical Research Council of Australia (APP1025350). The baseline
screening was supported by NHMRC grant (320860). We appreciate Dr Wendy Hoys
NHMRC Australia Research Fellowship (#511081).
Conceived and designed the experiments: OA WH ZW. Performed the experiments: OA WH
ZW. Analyzed the data: OA. Contributed reagents/materials/analysis tools: OA ZW WH.
Wrote the paper: OA ZW WH. Contributed to the study design: OA WH ZW. Involved in
obtaining the original data: WH ZW. Performed the literature searches, data analysis and drafted
the first version of the manuscript: OA. Revised and approved the final manuscript: OA WH
1. Diabetes Australia . Diabetes in Australia. 2013 Available: http://www.diabetesaustralia.com.au/ Understanding-Diabetes/Diabetes-in-Australia/
2. Wang Z , Hoy WE . Hypertension, dyslipidemia, body mass index, diabetes and smoking status in Aboriginal Australians in a remote community . Ethn Dis , 2003 ; 13 ( 3 ): 324 - 30 . PMID: 12894956
3. Daniel M , Rowley KG , McDermott R , Mylvaganam A , O'Dea K. Diabetes incidence in an Australian Aboriginal population. An 8-year follow-up study . Diabetes Care . 1999 ; 22 ( 12 ): 1993 - 1998 . PMID: 10587832
4. Australian Government . Aboriginal and Torres Strait Islanders and Diabetes. Action Plan . 2013 . Available: http://www.diabetesaustralia.com.au/Documents/DA/Media%20Releases/Aboriginal%20and% 20Torres%20Strait%20Islanders%20and%20Diabetes%20Action%20Plan .pdf
5. Shukla A. Type 2 Diabetes in Indigenous Communities: A Multifactorial Approach. Cross-sections . 2010 ; 6 : 59 - 71 .
6. Burke V , Zhao Y , Lee AH , Hunter E , Spargo RM , Gracey M , et al. Predictors of type 2 diabetes and diabetes-related hospitalisation in an Australian Aboriginal cohort . Diabetes Res Clin Pract . 2007 ; 78 ( 3 ): 360 - 368 . PMID: 17532084
7. Wang Z , Hoy WE , Si D. Incidence of type 2 diabetes in Aboriginal Australians: an 11-year prospective cohort study . BMC Public Health . 2010 ; 10 ( 487 ): 1 - 5 .
8. Baker IDI Heart & Biabetes Institure, Diabetes Australia . Diabetes the silent pandemic and its impact on Australia . 2012 . Available: http://www.diabetesaustralia.com.au/Documents/DA/What%27s%20New/ 12.03.14% 20Diabetes%20management%20booklet%20FINAL .pdf
9. Bjorntorp P. Metabolic implications of body fat distribution . Diabetes Care . 1991 ; 14 ( 12 ): 1132 - 1143 . PMID: 1773700
10. Bergman RN , Ader M. Free fatty acids and pathogenesis of type 2 diabetes mellitus . Trends Endocrinol Metab . 2000 ; 11 ( 9 ): 351 - 356 . PMID: 11042464
11. Kondalsamy-Chennakesavan S , Hoy WE , Wang Z , Briganti E , Polkinghorne K , Chadban S , et al. Anthropometric measurements of Australian Aboriginal adults living in remote areas: comparison with nationally representative findings . Am J Hum Biol . 2008 ; 20 ( 3 ): 317 - 324 . doi: 10.1002/ajhb.20729 PMID: 18203125
12. O'Dea K. Body fat distribution and health outcome in Australian Aborigines . Proc. Nutr. Soc. Aust . 1987 ; 12 : 56 - 65 .
13. Wang Z , Hoy WE . Body size measurements as predictors of type 2 diabetes in Aboriginal people . Int J Obes Relat Metab Disord . 2004 ; 28 ( 12 ): 1580 - 1584 . PMID: 15356663
14. Bambrick HJ . Relationships between BMI, waist circumference, hypertension and fasting glucose: Rethinking risk factors in Indigenous diabetes . Australian Indigenous Health Bulletin . 2005 ; 5 ( 4 ): 1 - 15 .
15. Schmidt MI , Duncan BB , Bang H , Pankow JS , Ballantyne CM , Golden SH , et al. Identifying individuals at high risk for diabetes: The Atherosclerosis Risk in Communities study . Diabetes Care . 2005 ; 28 ( 8 ): 2013 - 2018 . PMID: 16043747
16. Lindstrom J , Tuomilehto J. The diabetes risk score: a practical tool to predict type 2 diabetes risk . Diabetes Care . 2003 ; 26 ( 3 ): 725 - 731 . PMID: 12610029
17. Chien K , Cai T , Hsu H , Su T , Chang W , Chen M , et al. A prediction model for type 2 diabetes risk among Chinese people . Diabetologia . 2009 ; 52 ( 3 ): 443 - 450 . doi: 10.1007/s00125- 008 - 1232 - 4 PMID: 19057891
18. Chen L , Magliano DJ , Balkau B , Colagiuri S , Zimmet PZ , Tonkin AM , et al. AUSDRISK: an Australian Type 2 Diabetes Risk Assessment Tool based on demographic, lifestyle and simple anthropometric measures . Med J Aust . 2010 ; 192 ( 4 ): 197 - 202 . PMID: 20170456
19. Australian Government Department of Health. The Australian type 2 diabetes risk assessment tool (AUSDRISK): How do you score? 2010 . Available: http://www.diabetesaustralia. com.au/PageFiles/ 937/AUSDRISK%20Web% 2014 % 20July%2010 .pdf
20. Hoy WE , Baker PR , Kelly AM , Wang Z. Reducing premature death and renal failure in Australian aboriginals. A community-based cardiovascular and renal protective program . Med J Aust . 2000 ; 172 ( 10 ): 473 - 478 . PMID: 10901769
21. McDermott RA , Li M , Campbell SK . Incidence of type 2 diabetes in two Indigenous Australian populations: a 6-year follow-up study . Med J Aust . 2010 ; 192 ( 10 ): 562 - 565 . PMID: 20477730
22. Janiszewski PM , Janssen I , Ross R. Does waist circumference predict diabetes and cardiovascular disease beyond commonly evaluated cardiometabolic risk factors? Diabetes Care . 2007 ; 30 ( 12 ): 3105 - 3109 . PMID: 17712026
23. Wang Y , Rimm EB , Stanpfer MJ , Willett WC , Hu FB . Comparison of abdominal adiposity and overall obesity in predicting risk of type 2 diabetes among men . Am J Clin Nutr . 2005 ; 81 ( 3 ): 555 - 563 . PMID: 15755822
24. StataCorp , Stata Statistical Software: Release 12. College Station, TX: StataCorp LP . 2011 .
25. Heart Foundation . Absolute cardiovascular disease risk assessment . Heart Foundation Guide to management of hypertension . 2008 ; 1 - 6 .
26. Diabetes Australia. Type 2 Diabetes . 2013 . Available: http://www.diabetesaustralia. com.au/ Understanding-Diabetes/What-is-Diabetes/Type-2-Diabetes/.
27. Diabetes Australia Guideline Development Consortium. ( 2009 ). National evidence based guideline for case detection and diagnosis of type 2 diabetes . 2009 . 1 - 206 .
28. Chittleborough CR , Grant JF , Phillips PJ , Taylor AW . The increasing prevalence of diabetes in South Australia: The relationship with population ageing and obesity . Journal of the royal institute of public health . 2007 ; 121 : 92 - 99 .
29. Wang Z , Hoy WE . Waist circumference, body mass index, hip circumference and waist-to-hip ratio as predictors of cardiovascular disease in Aboriginal people . Eur J Clin Nutr . 2004 ; 58 ( 6 ): 888 - 93 . PMID: 15164109