Journal of the New Zealand Medical Association, 29-June-2007, Vol 120 No 1257
Ethnic differences in the prevalence of new and known diabetes mellitus, impaired glucose tolerance, and impaired fasting glucose. Diabetes Heart and Health Survey (DHAH) 2002–2003, Auckland New Zealand
Gerhard Sundborn, Patricia Metcalf, Robert Scragg, David Schaaf, Lorna Dyall, Dudley Gentles, Peter Black, Rodney Jackson
In 2003, one in 23 New Zealand (NZ) adults had self-reported diabetes mellitus.1 The prevalence of diabetes was significantly higher in Māori (males 9.5%, females 6.7%) and Pacific (males 8.0%, females 12.0%) populations [mostly of Samoan, Tongan, Niuean, or Cook Islands origin; hereafter termed ‘Pacific’] than in NZ Europeans (males 3.4%, females 2.4%).1 It has been suggested that for every person diagnosed with diabetes there is another person in the community undiagnosed.2
In November 2005, approximately 125,000 people had diagnosed diabetes; therefore according to this prediction the true diabetic population in NZ exceeds 250,000 people. It is predicted that more than 7,500 new people will be diagnosed with diabetes in 2006 and that more than 1,700 deaths will be attributable to diabetes.3
For Māori and Pacific, diabetes related-mortality rates are 10 times higher than for European people.4 A recent review of the epidemiology of diabetes in NZ lists the prevalence of known and undiagnosed diabetes from various NZ surveys and has called for a nationally agreed strategic plan on how to best monitor and control diabetes.5
Previous New Zealand studies and surveys that have been used to estimate the prevalence of diabetes have used self-report data or have been workforce surveys1,6–8 or are now more than a decade old.9 Self-report surveys are likely to underestimate diabetes prevalence as those who have diabetes and are yet to be diagnosed will be missed (under-reporting) and workforce surveys will be biased due to the ‘healthy worker’ effect. Moreover the majority of these surveys included insufficient samples of Māori and Pacific to allow for meaningful ethnic comparisons.
The South Auckland diabetes project (1991-4) is a useful comparison for this study as it was also a population based survey and had good numbers of both Māori and Pacific.10
The Diabetes Heart and Health (DHAH) survey has attempted to overcome these problems by using a population-based study design that included Glucose Tolerance Tests (GTT) for all non-diabetic participants. Targeted sampling of Māori and Pacific was undertaken to generate large representative samples of these communities.
The purpose of this study is to describe and compare ethnic differences in the prevalence of new and known diabetes mellitus, and impaired glucose tolerance and impaired fasting glucose levels.
The DHAH survey was a cross-sectional study that surveyed people aged 35–74 years, between January 2002 and December 2003. All participants were selected from within the Auckland region. Of the 4049 participants, 1014 were Māori, 1011 were Pacific, and 1745 were of European ethnicity.
Adults were recruited using two sampling frames: one was a cluster sample where random starting point addresses were obtained from Statistics New Zealand and the probability of selection was proportional to the number of people living in that mesh block (response rate 61.3%); and the other was a random sample taken from the November 2000 Auckland electoral rolls stratified into 5-year age bands and included all people living in the Auckland area, with the exception of the Franklin and Rodney electorates (response rate 65%). Participants were interviewed in places close to where they lived and all completed a self-administered questionnaire and a series of health measures. The 19 people who refused to have a GTT were excluded. Asians were also excluded.
Classification of ethnicity first gave priority to Māori ethnicity and followed an ‘ever-Māori’ approach used to improve undercounts in health data sets,11 followed by Pacific and Asian while all other participants formed the European comparison group, as used by Statistics New Zealand.12 Ethical approval was obtained from the Health and Disability Ethics Committees.
All participants received information in the mail with instructions not to eat any food from 10pm onwards the night before their survey was scheduled and to drink water only. Included in the information pack was a sterile urine container that was used to collect an early-morning urine sample (midstream). Most participants were contacted by phone prior to their survey appointment where instructions were explained again and any queries answered. On arrival at the survey location (from 8am to 10am) and after initial consent was given, a fasting blood sample was taken from all participants.
Participants were then asked whether they had been diagnosed with diabetes, and if so how old they were when they were first told, and what their current treatment was. Those who did not have previously diagnosed diabetes mellitus were then asked to complete a 2 hour Glucose Tolerance Test (GTT). This involved them having a drink consisting of 75g glucose after their initial blood test. A final blood test was then scheduled to be taken 2 hours after the first. During the wait-time, participants were asked to fill in other survey questionnaires and to not physically exert themselves in any way or consume any food or drink with the exception of water. Glucose samples were collected into fluoride tubes and stored on ice until taken to the lab for analyses.
Fasting blood samples were assayed using enzymatic methods, plasma glucose was measured using commercial reagents (Roche Products [NZ]), HbA1c was measured by high performance liquid chromatography, and micro-albumin was measured using an immunoturbidmetric method.
Categorisation of glucose tolerance status was evaluated by 1998 WHO criteria using fasting glucose ≥ 7.0 mmol/L or 2-hour post glucose load of ≥ 11.1 mmol/L for diabetes; fasting glucose < 7.0 mmol/L and 2-hour glucose between 7.8 and 11.0 mmol/L for Impaired Glucose Tolerance (IGT) and fasting glucose of 6.1- 6.9 mmol/L for Impaired Fasting Glucose (IFG). All participants were then classified as ‘known’ (from their past history), ‘new’-ly diagnosed, having ‘IGT’ or ‘IFG’ or ‘normal’ glucose functioning.
Leisure exercise was assessed using a three-month physical activity recall questionnaire.13 One question asked if participants had engaged in any vigorous activity at least once a week, in the past three months, long enough, that caused them to breathe hard or sweat. The other question asked if they had engaged in any moderate activity (that did not cause them to breathe hard or sweat).
Statistical analysis was undertaken using SAS version 9.1. Participant data were weighted according to the sampling frame that they were obtained from and means, standard errors and prevalence’s calculated using dual frame sampling methodology.14–16 SAS survey procedures (SURVEYMEANS, SURVEYREG, SURVEYFREQ AND SURVEYLOGISTIC) were used to calculate weighted means, adjusted means, percentages and odds ratios, respectively.17
The Rao-Scott modified Pearson Chi-squared test was used where appropriate with the reference category being the Europeans, because they constituted the largest sample. Odds Ratios were converted to Relative Risks as described by Zhang and Yu.18
The demographic characteristics are shown in Table 1. For the total study population, 48.0% were male; 27.0% aged <45 years, 26.4% aged 45–54 years, 24.4% aged 55–64 years, and 22.3% aged 65+ years; 26.9% of the participants were Māori, 26.8% were Pacific, and 46.3% were of European ethnicity.
The proportions of impaired fasting glucose (IFG), impaired glucose tolerance (IGT), newly diagnosed, previously diagnosed (known), and total diabetes are shown in Figure 1. Europeans had the lowest proportions, Pacific had the highest, and Māori were intermediate for prevalence of all diabetes states. Māori had 2.8 times higher and Pacific 4.1 times higher prevalence of total diabetes mellitus compared to Europeans.
Figure 1. Prevalence of diabetes states by ethnicity adjusted for age and sex
Lifestyle, socioeconomic status and demographic characteristics by diabetes status are presented in Table 2. These proportions have been adjusted for age, sex, and ethnicity. Those with abnormal diabetes status generally had significantly higher BMI, were older and exercised less. Compared to those with new or known diabetes and IGT categories, the IFG subgroup was not as distinctly differentiated from the ‘normal’ group.
Table 2. Percentage and mean demographic characteristics by diabetes status; adjusted for age, sex, and ethnicity
*0.01 <p< 0.05; ** 0.001<p<0.01; *** p<0.001 compared to ‘Normal’ group; †Not adjusted for sex; ‡Not adjusted for age; IGT: impaired glucose tolerance; IFG: impaired fasting glucose; Mod-ex: Participated in moderate exercise at least 1 × per week in past 3 months; Vig-ex: Participated in vigorous exercise at least 1 x per week in past 3 months.; Good+ health: Rated personal health as good or better.
Table 3 compares the risk of having newly diagnosed or known diabetes by ethnic group. Māori aged 45–54 and 55–64 years were found to have significantly higher risk of known diabetes compared to Europeans (RR: 6.4 and 4.1 respectively). For Pacific, all age groups had a significantly higher risk of known diabetes than Europeans, with the highest being in the 55–64 year age group (RR: 9.3).
Only Pacific participants in the <45, and 45–55 year age groups were found to have a significantly higher risk of new diabetes status (RR: 11.6 and 4.2 respectively) compared to Europeans.
Table 3. Relative risk (RR) of new and known diabetes by ethnic group, adjusted for sex
*0.01 <p< 0.05; **0.001<p<0.01; ***p<0.001 compared to Europeans.
The highest proportion of new diabetes was observed in the Pacific 45-54 age group of 6.9% (Europeans: 2.2%, Māori: 2.5%). The largest proportion of previously diagnosed (known) diabetes was also reported by the Pacific ethnic group, aged 55-64 of 38.2%. This is compared to Europeans 4.4%, and Māori 17.4%.
The only significant difference in IGT was observed in the Pacific ethnic group aged 45-54 years compared to the Europeans. For IFG the only significant differences were observed in Māori aged 45-54, and Pacific aged <45 and 45-54 years. Generally a clear trend was observed where Pacific had the highest risk and Māori intermediate for both IGT and IFG compared to Europeans. However two exceptions to this trend existed. The 55-64 IFG Māori group and the 65+ IFG/IGT Pacific groups reported lower (but not significantly different) risks compared to Europeans.
Table 4. Relative risk (RR) of impaired glucose tolerance (IGT) and impaired fasting glucose (IFG) by age group and ethnicity, adjusted for sex
*0.01 <p< 0.05; ** 0.001<p<0.01; *** p<0.001 compared to Europeans.
Figure 2 shows the cumulative proportion of newly diagnosed diabetes mellitus by age group and ethnicity. This graph shows that Māori follow a similar trend to Europeans in spite of having a higher prevalence of both new and known diabetes mellitus. The Pacific ethnic group however follow a clearly different path. For Pacific, the same proportion of newly diagnosed participants were diagnosed approximately ten years earlier compared to both Māori and Europeans. For example 80% of newly diagnosed Pacific participants had been diagnosed by the 45-54 age group. This occurred closer to the 55-64 age-group for both Māori and Europeans.
Figure 2. Cumulative proportion of new diabetes status by age and ethnicity
Table 5 presents the findings of 2 multivariate models for newly diagnosed diabetes mellitus. Model 1 shows that being Māori or Pacific increases the odds of being newly diagnosed with diabetes mellitus compared with Europeans. However, only Pacific ethnicity was statistically significant. After adjustment for Body Mass Index (BMI) in Model 2 both odds ratios for Māori and Pacific ethnicity dropped making Pacific ethnicity now insignificant.
Table 5. Multivariate odds ratios (95% CI) for ‘newly’ diagnosed diabetes mellitus
Model 1 includes age, male, Māori and Pacific; Model 2 includes Model 1 plus BMI.
Table 6 shows 2 multivariate models for previously diagnosed (known) diabetes mellitus. Model 1 show 4 times higher odds for Māori and 6.6 times higher odds for Pacific of previously diagnosed diabetes compared to Europeans. Adjustment for BMI in Model 2 saw a reduction of the odds in both Māori and Pacific compared to Europeans, but did not eliminate ethnic differences in people with previously diagnosed diabetes mellitus.
Table 6. Multivariate odds ratios (95% CI) for ‘known’ diabetes mellitus
Model 1 includes age, male, Māori and Pacific; Model 2 includes Model 1 plus BMI.
It has been frequently reported that between a third and a half of all diabetes in the community remains undiagnosed2 and that this may be experienced more by Pacific people.19 The common mantra that ‘for every known case of diabetes there is another undiagnosed in the community’, is not supported by our study and has significant public health implications in estimating the projected burden of undiagnosed diabetes in New Zealand by ethnic group.
In the current study, the proportion of Europeans that were newly diagnosed with diabetes was equal to 46% of those with known diabetes. This suggests that for every two European people with previously diagnosed diabetes there approximately one (0.92) person in the community undiagnosed.
For Māori and Pacific, the proportions that were newly diagnosed with diabetes were equal to 32% and 21% respectively, of those with known diabetes. This suggests that for every three Māori and every five Pacific people with previously diagnosed diabetes there would one person in the community undiagnosed (0.96 for Māori, 1.05 for Pacific).
In contrast, a study that measured new and known diabetes in adults aged 40–70 years in South Auckland during 1991-94, found that the proportion of new diabetes in Europeans was equal to 52% of known. For Māori and Pacific, these proportions were 77% and 81% respectively.10 These findings suggest that diabetes screening for Māori and Pacific have improved considerably over the past decade.
Prevalences of previously diagnosed diabetes found in this study had similar ethnic patterns to those reported in the 2002/03 NZ Health Survey. However the self-reported prevalences of previously diagnosed diabetes tended to be considerably lower in the NZ Health Survey. In absolute terms, the DHAH survey prevalences were higher than the NZ Health Survey data by approximately 1.4% for Europeans, 4.0% for Māori, and 9.5% for Pacific ethnic groups. These differences are in part due to the differing age structure of each survey. The NZ Health survey sampled from 15 years and above compared to 35 years for the DHAH. These surveys were both conducted during 2002/03.
In contrast to a cross-sectional survey carried out in South Auckland from 1992-199511 and the Workforce Survey in 1988-1990,20 Pacific people now have a poorer profile than the Māori population. The South Auckland study reported age adjusted rates of known diabetes of 5.2% for Europeans, 7.3% for Māori and 6.0% for Pacific peoples compared to 3.9%, 12.0%, and 19.5%, respectively, in the DHAH.
For those with previously diagnosed (known) diabetes, Pacific aged 55-64 reported the largest relative risk (RR: 9.33 compared to Europeans), this compares to Pacific aged 50-54 (RR: 11.8) from the Workforce Survey. Prevalences of known diabetes by age group and ethnicity followed expected trends with Pacific people having the highest prevalences and Māori intermediate.
However the patterns in RR were more mixed for new diabetes. Māori had the highest RR for new diabetes in the 55-64 and 65+ age groups with Pacific intermediate. For the <45 and 45-54 age groups Pacific had the highest risk. This could be due to both earlier onset of diabetes in Pacific and less robust screening for diabetes in Māori compared to Pacific in the younger age groups. Risk of new diabetes in Māori ranged from 1.38–4.97 compared to Europeans and for Pacific they ranged from 0.99–11.61. The largest risk for new diabetes was found in the Pacific age group of <45 years (RR: 11.61) which was also the case for the Workforce Survey (RR: 9.5).
The Workforce Survey conducted during 1988-9021 reported newly diagnosed diabetes in 1.7% of Europeans, 9.7% Māori, and 7.7% Pacific people.22 However, these were likely to be lower than the general population as they were employees (healthy worker effect). Prevalences found in the DHAH were 1.8% for Europeans, 3.8% Māori, and 4.0% Pacific.
The marked decrease in Māori of newly diagnosed diabetes is interesting and may be the result of improved health and improved access to healthcare with the emergence of many new Māori healthcare providers over the past two decades, resulting in earlier detection of diabetes. Measurement and classification of Māori ethnicity could also have influenced this difference. This study used an ‘ever-Māori’ approach to assign ethnicity which decreases the likelihood of under-reporting.11
A limitation of this study is that using Electoral role based and cluster sampling frames did not allow for ethnic specific response rates to be determined. Although the overall response rate was not as high as in previous Auckland risk factor studies, it has been shown in the Atherosclerosis Risk in Communities Study22 that response rates lower than those in our study produced relatively small errors in the estimates of prevalence of common cardiovascular disease risk factors.
Participants with known diabetes were more likely to engage in moderate exercise when compared to those newly diagnosed. This suggests that once a diagnosis is made increased physical activity may have been recommended to these people.
Having received further tertiary education had a protective association with diabetes. Levels of tertiary education were lower for all categories of impaired glucose tolerance when compared to the ‘normal’ reference group. (Table 2) This may be due to education leading towards higher socio-economic status, and also an increased awareness of healthy lifestyles, diabetes risk factors and symptoms.
The difference observed in the cumulative proportions of new diabetes between Pacific and non-Pacific ethnic groups (Figure 2) showed that a larger proportion of Pacific people generally experience earlier onset of diabetes. This difference equates to Pacific people being diagnosed up to 10 years earlier than Europeans and suggests that Pacific people will live with diabetes and its complications significantly longer and/or have earlier mortality. This figure could also imply that the Māori population may follow a more similar disease profile/pattern to Europeans than Pacific.
It is important to note that the age at diagnosis is not necessarily the age of development of diabetes, and that the time between development and diagnosis may vary between ethnic groups.
BMI and age were found to be the most significant factors associated with people newly diagnosed with diabetes, IGT, and IFG (Table 2, 5) which supports the focus that many health campaigns have on prevention and control of obesity to lower the prevalence of diabetes. Model 2 from Table 5, showed that adjusting for BMI alone reduced ethnic differences in new diabetes prevalence. However, although adjustment for BMI in people previously diagnosed with diabetes mellitus reduced the odds in both Māori and Pacific people compared to Europeans, it did not eliminate these ethnic differences (Table 6).
Increasing and maintaining health promotion programmes centred on living healthy lifestyles (nutrition and activity) to keep a healthy BMI will continue to be the most appropriate method to prevent and manage diabetes in New Zealand.
Competing interests: None.
Author information: Gerhard McDonald-Sundborn, Research Fellow in Pacific Health; Patricia Metcalf, Senior Lecturer in Biostatistics; Robert Scragg, Associate Professor of Epidemiology; David Schaaf, Senior Research Fellow in Pacific Health; Lorna Dyall, Senior Lecturer in Māori Health; Dudley Gentles, Research Fellow in Māori Health; Peter Black, Associate Professor of Medicine; Rodney Jackson, Professor of Epidemiology, Section of Epidemiology and Biostatistics, School of Population Health
University of Auckland, Auckland
Acknowledgments: This research was funded by the Health Research Council of New Zealand and was carried out in the Section of Epidemiology and Biostatistics/Section of Pacific Health, School of Population Health, University of Auckland.
We also thank the participants who took part in this survey: Wayne and Sola McDonald-Sundborn and Bruno and Vera Zarins for their helpful discussions, comments, and feedback.
Correspondence: Gerhard Sundborn, Section of Epidemiology and Biostatistics, School of Population Health, University of Auckland, Private Bag 92019, Auckland 1. Fax: (09) 3737 503; email: firstname.lastname@example.org
This article was corrected on 26 October 2007 to reflect the Erratum at http://www.nzma.org.nz/journal/120-1264/2797
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads)
Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals