Epidemiology/Health Services/Psychosocial Research O R I G I N A L
A R T I C L E
The Prevalence of Diabetes in the Kingdom of Tonga STEPHEN COLAGIURI, MD1 RUTH COLAGIURI, BED1 SIVA NA’ATI, DIPDIET2
SOANA MUIMUIHEATA, DIPDIET2 ZAFIRUL HUSSAIN, MSC1 TANIELA PALU, MD2
OBJECTIVE — To determine the prevalence of diabetes, impaired glucose metabolism, and related risk factors in Tonga. RESEARCH DESIGN AND METHODS — A randomly selected representative national sample of 1,024 people aged ⬎15 years was surveyed. Each participant had fasting blood glucose and HbA1c measured. Subjects with a fasting blood glucose ⬎5.0 mmol/l (90 mg/dl) and ⬍11.1 mmol/l (200 mg/dl) or a fasting blood glucose ⱕ5.0 mmol/l and an HbA1c ⬎6.0% and every fifth subject with a fasting blood glucose ⱕ5.0 mmol/l and a normal HbA1c had a 75-g oral glucose tolerance test (OGTT). A total of 472 individuals had an OGTT based on these criteria. Subjects with a fasting blood glucose ⱖ11.1 mmol/l and an elevated HbA1c were diagnosed as having diabetes. RESULTS — The mean age was 41.3 years, and the mean BMI was 32.3 kg/m2. The agestandardized prevalence of diabetes was 15.1% (CI 12.5–17.6), 12.2% (8.7–15.8) in men and 17.6% (14.0 –21.1) in women (NS), of which only 2.1% was previously diagnosed. A total of 75% of people with newly diagnosed diabetes had a fasting plasma glucose ⱖ7.0 mmol/l (126 mg/dl). The prevalence of impaired glucose tolerance was 9.4% (7.3–11.5) and of impaired fasting glycemia 1.6% (0.7–2.6). Undiagnosed diabetes was significantly associated with increasing age, obesity, hypertension, and a family history of diabetes. CONCLUSIONS — The current prevalence of diabetes in Tonga is 15.1%, of which 80% is undiagnosed. A similar survey in 1973 reported a 7.5% diabetes prevalence, indicating a doubling of diabetes over the past 25 years. In addition, lesser degrees of glucose intolerance are common, and much of the community is overweight Diabetes Care 25:1378 –1383, 2002
N
oncommunicable disease associated with changes in lifestyle and diet has become a major public health problem in developing countries. Diabetes is at the forefront of noncommunicable diseases (1–3), and a number of studies have documented that the prevalence of diabetes in the Pacific Islands is generally higher than in developed countries but shows a wide variation, with Nauru having the highest prevalence
(40%) and with some rural areas of Papua New Guinea having the lowest, with virtually no diabetes (1–10). The Kingdom of Tonga is one of many Pacific Islands experiencing a noticeable change in disease pattern and the emergence of noncommunicable diseases as a major threat to the health of the population. In 1997, a diabetes and related noncommunicable diseases control program was initiated in the Kingdom of Tonga. To
● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●
From the 1Australian Centre for Diabetes Strategies, Randwick, Australia; and the 2Diabetes Centre, Vaiola Hospital, Tonga. Address correspondence and reprint requests to Professor Stephen Colagiuri, Australian Centre for Diabetes Strategies, Prince of Wales Hospital, High St., Randwick 2031, NSW, Australia. E-mail:
[email protected]. Received for publication 8 November 2001 and accepted in revised form 30 April 2002. Abbreviations: ADA, American Diabetes Association; dBP, diastolic blood pressure; FcapBG, fasting capillary blood glucose; IFG, impaired fasting glycemia; IGT, impaired glucose tolerance; OGTT, oral glucose tolerance test; sBP, systolic blood pressure; WHR, waist-to-hip ratio. A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion factors for many substances.
1378
allow future evaluation of the impact of the program, a baseline prevalence study for diabetes and related risk factors was conducted between 1998 and 2000. This paper reports the prevalence of type 2 diabetes, impaired glucose metabolism, and related risk factors. RESEARCH DESIGN AND METHODS — The Kingdom of Tonga is situated east of Fiji and south of Samoa and scattered between latitudes 15° and 23°S and longitudes 173° and 176°W. The Tongan people are polynesian, and the islands are believed to have first been inhabited during the 10th to 15th centuries (11). The first documented contact with Europeans was by the Dutch navigators in 1616. Tonga consists of 169 lowlying coral and volcanic islands of which 37 are inhabited. There are three main island groups—Tongatapu, on which the capital Nuku’alofa is situated, Vava’u, and Ha’apai. The 1996 census documented a total population of 97,784 with 59,526 people aged 15 years and older (12). Sampling Procedure The survey was conducted in two parts, Tongatapu in 1998 and Vava’u and Ha’apai in 2000. The overall sampling procedure was designed in collaboration with the Department of Statistics, Government of Tonga. A multi-stage cluster sampling design was used to select a representative sample of the Tongan population. In Tongatapu census blocks were randomly selected, and in Vava’u and Ha’apai villages were randomly selected. Every fourth household was selected systematically from each census block or village, and to reflect the age and sex distribution of the Tongan population, a designated male or female ⬎15 years of age was invited to participate. On the island of Tongatapu, 832 subjects were selected for participation in the study. A total of 66 individuals were excluded from participation, 16 had known type 2 diabetes (confirmed self-reported or taking diabetes medication), and others were either ill, pregnant, or were nonTongans. Among the eligible subjects, 45 refused to participate, 61 consented but DIABETES CARE, VOLUME 25, NUMBER 8, AUGUST 2002
Colagiuri and Associates
Table 1—Details of survey participants Men
n Age (years) BMI (kg/m2) Weight (kg) Waist circumference (cm) WHR sBP (mmHg) dBP (mmHg) Family history of diabetes (%) Cigarette smoking (%) Alcohol (%) Traditional medicine (%) Creatinine (mmol/l) Total cholesterol (mmol/l) HDL cholesterol (mmol/l) Triglycerides (mmol/l) Calculated LDL (mmol/l)* Total cholesterol-to-HDL ratio Urine albumin-to-creatinine ratio (mg/mmol)
1998 Survey
2000 Survey
Total
P
271 41.0 ⫾ 15.2 29.7 ⫾ 5.3 91.9 ⫾ 17.5 98.2 ⫾ 13.6 0.89 ⫾ 0.07 134.9 ⫾ 19.6 82.2 ⫾ 11.9 28.3 53.1 33.5 56.8 0.09 ⫾ 0.02 5.32 ⫾ 1.12 1.11 ⫾ 0.27 1.60 ⫾ 1.24 3.54 ⫾ 0.99 5.01 ⫾ 1.39 4.35 ⫾ 36.39
162 47.3 ⫾ 13.9 31.0 ⫾ 5.4 95.5 ⫾ 17.1 100.8 ⫾ 12.9 0.92 ⫾ 0.08 130.5 ⫾ 16.6 80.3 ⫾ 9.8 22.2 52.5 13.0 59.3 0.10 ⫾ 0.02 4.95 ⫾ 1.08 0.99 ⫾ 0.23 1.31 ⫾ 0.78 3.37 ⫾ 0.93 5.24 ⫾ 1.44 1.43 ⫾ 2.68
433 43.4 ⫾ 15.0 30.2 ⫾ 5.4 93.2 ⫾ 17.4 99.2 ⫾ 13.4 0.90 ⫾ 0.07 133.2 ⫾ 18.6 81.5 ⫾ 11.2 26.0 52.9 25.8 57.7 0.09 ⫾ 0.02 5.18 ⫾ 1.12 1.06 ⫾ 0.26 1.49 ⫾ 1.10 3.48 ⫾ 0.97 5.10 ⫾ 1.41 3.26 ⫾ 28.88
0.000 0.016 0.038 0.052 0.000 0.016 0.095 0.167 0.893 0.000 0.620 0.001 0.001 0.000 0.006 0.089 0.114 0.310
Data are means ⫾ SD unless otherwise indicated. *Sample size for LDL cholesterol—1998 survey: men 253, women 330; 2000 survey: men 160, women 250.
failed to attend, and 50 were unavailable at the time of survey, giving an overall response rate of 80%. Two subjects who were tested were not included in the final analysis because of missing information, giving a final number of 608 people. In Vava’u and Ha’apai 527 people were invited to participate, of whom 25 were ineligible (10 with known type 2 diabetes), 20 refused to participate, 22 agreed but did not attend, and 44 were unavailable. In total, 416 people participated (202 in Vava’u and 214 in Ha’apai), giving a response rate of 83%. The total survey population was 1,024. Survey procedure Participants were given appointments to attend the survey site between 0800 and 1000 h after an overnight fast and were asked to bring an early morning urine sample. On arrival, trained interviewers administered a questionnaire that collected demographic details and information on family history of diabetes, personal history of illness and lifestylerelated risk factors, and use of traditional medicine. Height (to the nearest millimeter) and weight (in kilograms) were measured without shoes and in light clothing, and BMI was calculated. Waist and hip circumferences were measured, and waist-to-hip ratio (WHR) was calculated. DIABETES CARE, VOLUME 25, NUMBER 8, AUGUST 2002
High WHR was defined as ⬎1.00 for men and ⬎0.85 for women. Normal waist circumference was defined as ⬍94 cm for men and ⬍80 cm for women. Men with waist circumference ⱖ94 and ⬍102 cm and women with waist circumference ⱖ80 and ⬍88 cm were defined as Level 1. Men with waist circumference ⱖ102 cm and women with waist circumference ⱖ88 cm were defined as Level 2 (13). Blood pressure was measured to the nearest 2 mmHg while in the sitting position in the right arm after 10 min rest. Systolic blood pressure (sBP) was recorded at the level of appearance of sound and diastolic blood pressure (dBP) at the level of sound disappearance (phase V). Hypertension was defined as sBP ⱖ140 mmHg and/or dBP ⱖ90 mmHg or the use of antihypertensive medication. Urine was analyzed for albuminuria, which was expressed as albumin-to-creatinine ratio with the normal level defined as a ratio ⬍2.5 mg/mmol for men and ⬍3.5 mg/ mmol for women. Fasting venous blood was collected for measurement of plasma glucose and lipids including total cholesterol, HDL cholesterol, and triglycerides. At the same time, each subject had a fasting capillary whole blood glucose level measured by finger prick using a portable Hemocue meter (HemoCue AB, Angelholm, Sweden), and HbA1c was mea-
sured using the DCA 2000 analyzer (Bayer, Elkhart, IN) (normal range 4 – 6%). All subjects with a fasting capillary whole blood glucose ⬎5.0 and ⬍11.1 mmol/l underwent an oral glucose tolerance test (OGTT). Subjects with a fasting blood glucose ⱕ5.0 mmol/l and an HbA1c ⬎6.0% also had an OGTT. Every fifth subject having a fasting blood glucose ⱕ5.0 mmol/l and a normal HbA1c level also had an OGTT. A total of 472 individuals had an OGTT based on these criteria. Subjects with a fasting blood glucose ⱖ11.1mmol/l and an elevated HbA1c level were diagnosed as having diabetes. A standard OGTT was performed with a fasting venous blood collection for measurement of plasma glucose, ingestion of 75 g glucose monohydrate dissolved in 250 ml water, and collection of another blood sample 2 h later. All venous blood samples were promptly centrifuged, stored at ⫺20°C, and air transported to the Prince of Wales Hospital in Sydney for analysis. Blood sample assays were performed using a Beckman LX20 analyzer. Glucose was determined using the hexokinase method, total cholesterol using cholesterol esterase, oxidase, and peroxidase to produce a quinoneimmine product, HDL cholesterol after solubilizing with detergent, triglyceride after hydroly1379
Diabetes prevalence in Tonga
Table 1—Continued Women 1998 Survey
2000 Survey
Total
P
Men and women combined total
337 38.1 ⫾ 14.2 33.2 ⫾ 6.0 91.2 ⫾ 17.3 98.9 ⫾ 12.7 0.83 ⫾ 0.06 127.8 ⫾ 21.0 80.0 ⫾ 11.7 30.9 14.0 6.3 69.4 0.07 ⫾ 0.01 5.02 ⫾ 1.10 1.15 ⫾ 0.25 1.15 ⫾ 0.75 3.36 ⫾ 0.95 4.53 ⫾ 1.24 1.23 ⫾ 1.63
254 42.1 ⫾ 12.2 34.5 ⫾ 6.3 95.5 ⫾ 18.3 100.9 ⫾ 13.9 0.85 ⫾ 0.08 125.4 ⫾ 16.9 79.4 ⫾ 9.8 32.1 5.9 0 63.0 0.08 ⫾ 0.01 4.75 ⫾ 0.97 1.06 ⫾ 0.26 1.07 ⫾ 0.68 3.20 ⫾ 0.86 4.65 ⫾ 1.29 1.47 ⫾ 2.31
591 39.8 ⫾ 13.5 33.8 ⫾ 6.2 93.0 ⫾ 17.9 99.7 ⫾ 13.3 0.84 ⫾ 0.07 126.8 ⫾ 19.4 79.8 ⫾ 10.9 31.4 10.5 3.6 66.7 0.07 ⫾ 0.01 4.90 ⫾ 1.05 1.11 ⫾ 0.26 1.12 ⫾ 0.72 3.29 ⫾ 0.92 4.58 ⫾ 1.27 1.33 ⫾ 1.95
0.000 0.007 0.004 0.060 0.008 0.124 0.508 0.740 0.002 0.000 0.100 0.000 0.001 0.000 0.184 0.044 0.289 0.143
1024 41.3 ⫾ 14.3 32.3 ⫾ 6.1 93.1 ⫾ 17.6 99.5 ⫾ 13.3 0.87 ⫾ 0.08 129.5 ⫾ 19.3 80.5 ⫾ 11.0 29.1 28.5 12.9 62.9 0.08 ⫾ 0.02 5.02 ⫾ 1.09 1.09 ⫾ 0.26 1.28 ⫾ 0.92 3.37 ⫾ 0.94 4.80 ⫾ 1.35 2.15 ⫾ 18.86
sis to glycerol by lipase, and creatinine after reacting with alkaline picrate. Urine albumin was measured by the immunoturbidimetry method. LDL cholesterol was calculated as follows: total cholesterol ⫺ triglycerides/ 2.2 ⫺ HDL cholesterol
provided that triglycerides were ⬍4.0 mmol/l. Diabetes, impaired fasting glycemia (IFG), and impaired glucose tolerance (IGT) were defined according to the new diagnostic criteria (14,15). Statistical analysis Statistical analysis was performed using SPSS statistical software release 10.0. Descriptive statistics of the study subjects were computed as unadjusted means ⫾ SD and compared using Student’s t test for continuous variables and 2 test for categorical variables. Age-specific and age-adjusted prevalence and the 95% CIs were calculated against the 1996 Tongan census data. To enable comparison with world populations, age-standardized prevalence rates for the Segi population aged 30 – 64 years were also calculated. Association with a binary dependent variable was examined by multiple logistic regression analysis. To evaluate the asso1380
ciation of diabetes status and other risk factors, glucose status was dichotomized into two categorical variables—normal glucose tolerance and type 2 diabetes. People with IGT and/or IFG were excluded from this analysis. Age (continuous), sex, BMI, waist circumference, WHR, family history of diabetes, hypertension, and lipid fractions were considered as independent variables. A twotailed P value ⬍0.05 was considered significant. RESULTS — Details of the participants are shown in Table 1. For the group (mean ⫾ SD), age was 41.3 ⫾ 14.3 years, BMI 32.3 ⫾ 6.1 kg/m2, waist circumference 99.5 ⫾ 13.3 cm, and WHR 0.87 ⫾ 0.08. Compared with the Tongatapu group (1998 Survey), the Vava’u/Ha’apai group (2000 Survey) were significantly older (44.1 vs. 39.4 years; P ⬍ 0.001) and heavier (BMI 33.2 vs. 31.6 kg/m2; P ⬍ 0.001) but had lower sBP (127.3 vs. 131 mmHg; P ⫽ 0.003), lower total cholesterol (4.8 vs. 5.2 mmol/l; P ⬍ 0.001), lower HDL cholesterol (1.03 vs. 1.13 mmol/l; P ⬍ 0.01), and lower triglycerides (1.16 vs. 1.35 mmol/l; P ⫽ 0.001) and fewer were smokers (24.1 vs. 31.5%; P ⫽ 0.01) or consumed alcohol (5 vs. 18.3%; P ⬍ 0.001). The prevalence of undiagnosed type 2
diabetes and glucose intolerance are shown in Table 2. The overall crude prevalence of undiagnosed diabetes was 10.4% (CI 8.5–12.3), 9.5% (6.7–12.3) in men and 11.0% (8.5–13.5) in women (NS). Prevalence was slightly but not significantly higher in Vava’u/Ha’apai compared with Tongatapu (11.3% [CI 8.3– 14.3] vs. 9.7% [7.3–12.1]). The prevalence of diagnosed type 2 diabetes was 2.1% (2.2% in Tongatapu and 2.0% in Vava’u/Ha’apai), giving an overall agestandardized prevalence (30 – 64 years, Tongan 1996 census) of type 2 diabetes of 15.1% (12.5–17.6), 12.2% (8.7–15.8) for men and 17.6% (14.0 –21.1) for women (NS). Standardized for the Segi world population aged 30 – 64 years, the prevalence of type 2 diabetes was 16.0% (13.3– 18.6). The age-specific prevalence rates are presented in Fig. 1 and show rates of type 2 diabetes of ⬃20% from age 50 years. The overall age-standardized prevalence rate (30 – 64 years) for IGT was 9.4% (CI 7.3–11.5), 8.7% (5.7–11.8) in men and 9.8% (6.8 –12.7) in women (Table 2). IGT was more common in Vava’u/ Ha’apai than in Tongatapu (11.4 vs. 8.1%), but this difference was not statistically significant. Prevalence rates for IFG were 1.6% (0.7–2.6) overall, 2.4% (0.7– 4.1) in men, 1.1% (0.1–2.1) in women, DIABETES CARE, VOLUME 25, NUMBER 8, AUGUST 2002
Colagiuri and Associates
Table 2—Age-specific and age-standardized prevalence of undiagnosed type 2 diabetes and impaired glucose metabolism by sex Men n
Type 2 diabetes
IGT
Women IFG
n
Type 2 diabetes
IGT
Men and women combined IFG
Type 2 diabetes
IGT
IFG
Age group (years) 15–19 20 0 0 0 39 0 0 0 0 0 0 20–24 31 0 0 3.2 40 0 2.5 0 0 1.4 1.4 25–29 31 0 0 0 65 0 3.1 0 0 2.1 0 30–34 58 6.9 3.4 3.4 84 6.0 10.7 1.2 6.3 7.7 2.1 35–39 54 5.6 5.6 1.9 82 9.8 3.7 0 8.1 4.4 0.7 40–44 45 15.6 11.1 6.7 70 14.3 8.6 1.4 14.8 9.6 3.5 45–49 43 7.0 14.0 0 70 12.9 14.3 2.9 10.6 14.2 1.8 50–54 36 13.9 8.3 0 52 25.0 9.6 1.9 20.5 9.1 1.1 55–59 38 13.2 13.2 2.6 33 18.2 12.1 0 15.5 12.7 1.4 60–64 34 17.6 11.8 0 25 32.0 12.0 0 23.7 11.9 0 ⱖ65 43 18.6 11.6 2.3 31 19.4 12.9 0 18.9 12.2 1.4 Total 433 9.5 7.6 2.1 591 11.0 8.0 0.8 10.4 7.8 1.4 Age-standardized* 6.22 4.83 1.74 8.66 6.42 0.51 7.46 5.78 1.07 prevalence ⱖ15 years 95% CI 4.43–8.00 3.25–6.41 0.47–3.00 6.64–10.69 4.51–8.34 0.07–0.96 6.12–8.80 4.50–7.06 0.45–1.69 Age-standardized* 10.55 8.72 2.39 14.95 9.76 1.10 12.86 9.37 1.64 prevalence 30–64 years 95% CI 7.16–13.94 5.65–11.80 0.65–4.13 11.47–18.43 6.84–12.67 0.14–2.06 10.43–15.29 7.25–11.50 0.72–2.56 Data are n or %. *Age and sex standardized to 1996 Tongan census.
1.0% (CI 0.0 –1.9) in Tongatapu, and 2.4% (0.7– 4.0) in Vava’u/Ha’apai (NS). A total of 241 people (20% of the total population) eligible for the survey did not participate, of whom 94 could not be located at the selected household, mainly because they were overseas or at another
island. Data on age and sex was available for 141 of the 148 people who refused to participate or who agreed but did not attend for the survey. The mean ⫾ SD age of this group was 39.3 years, not significantly different to the study cohort. However, there were twice as many men than
Figure 1—Prevalence of diabetes (known and newly diagnosed) in Tongan men and women by age group.
DIABETES CARE, VOLUME 25, NUMBER 8, AUGUST 2002
women in the nonparticipating group (96 vs. 45), mainly related to more men being unwilling to take time off work to attend the survey. Table 3 shows the association between risk factors and undiagnosed type 2 diabetes in a multiple logistic regression model compared with people with normal glucose tolerance. Increasing age, abnormal WHR, the presence of hypertension, a family history of diabetes, and microalbuminuria were all significantly associated with undiagnosed type 2 diabetes. However, none of the lipid levels, total cholesterol, HDL cholesterol, triglycerides, calculated LDL cholesterol, or total-to-HDL cholesterol ratio was significantly associated. Cardiovascular risk factors were common in survey participants. Of the total population, 60.3% had a BMI ⱖ30 kg/ m2, 65.1% had an increased waist circumference, 37.3% had hypertension, 33% had a total cholesterol ⱖ5.5 mmol/l, and 30.4% had an HDL cholesterol ⬍1.0 mmol/l. All risk factors, except some indexes of overweight, were significantly more common in men. There was a highly significant linear correlation between simultaneously mea1381
Diabetes prevalence in Tonga
Table 3—Multiple logistic regression model of variables associated with type 2 diabetes
Variables Age (years) Sex Female Male BMI (kg/m2) ⬍25.0 25.0–29.9 ⱖ30.0 WHR Normal Higher Hypertension No Yes Family history of diabetes No Yes Microalbumin Normal Abnormal
Logistic regression coefficient
Odds ratio
95% CI
P
0.040
1.04
1.02–1.06
⬍0.001
0.124
1.00* 1.13
0.61–2.11
0.695
⫺0.145 0.307
1.00* 0.87 1.36
0.21–3.61 0.31–6.05
0.843 0.687
1.075
1.00* 2.93
1.63–5.26
⬍0.001
1.153
1.00* 3.17
1.90–5.28
⬍0.001
0.886
1.00* 2.43
1.50–3.93
⬍0.001
1.182
1.00* 3.26
1.71–6.22
⬍0.001
*Reference category: multiple logistic regression model includes age, sex, BMI, waist circumference, WHR, family history of diabetes, hypertension, total cholesterol, HDL cholesterol, triglycerides, total-to-HDL cholesterol ratio, and microalbumin as risk factors. Statistically significant variables shown, except for sex and BMI. ⫺2 Log likelihood: 488.61; model 2 statistic: 155.52 (P ⬍ 0.001); dependent variable: normal glucose tolerance ⫽ 0 and type 2 diabetes ⫽ 1; normal WHR: men ⬍0.95 and women ⬍0.80; high WHR: men ⱖ0.95 and women ⱖ0.80; hypertension: sBP ⱖ140 mmHg and/or dBP ⱖ90 mmHg or hypertensive medication; microalbuminuria: men ⬎2.5 mg/mmol and women ⬎3.5 mg/mmol.
sured finger-prick fasting capillary blood glucose (FcapBG) and laboratory fasting venous plasma glucose (mean 5.37 vs. 5.38 mmol/l, respectively; correlation coefficient 0.961). The decision to perform an OGTT was based on the FcapBG and HbA1c results. The majority of those having an OGTT had an FcapBG ⬎5.0 mmol/l and ⬍11.1 mmol/l (n ⫽ 391), whereas 3 subjects had an FcapBG ⱕ5.0 mmol/l and an HbA1c ⬎6.0%. A total of 78 subjects with an FcapBG ⱕ5 mmol/l and a normal HbA1c had an OGTT. None had diabetes, but 6 of the 78 (7.5%) met the diagnostic criteria for IGT. The 2-h plasma glucose levels in these individuals ranged from 7.8 to 9.6 mmol/l, with a mean of 8.6 mmol/l. Of the 106 subjects with newly diagnosed diabetes based on the results of the OGTT (15), 79 (75%) had a fasting plasma glucose ⱖ7.0 mmol/l and would have met the American Diabetes Association (ADA) diagnostic criteria for diabetes (14). Of the remaining 27 individuals, 17 (16% of all people with newly diagnosed diabetes) met the ADA criteria for IFG, 1382
and the other 10 (9%) had a fasting plasma glucose ⱕ6.0 mmol/l and would have been designated as having normal glucose tolerance (14). CONCLUSIONS — T h i s s u r v e y found a high prevalence of diabetes in Tonga with an age- and sex-standardized prevalence of 15.1% (CI 12.5–17.6). Rates in women were higher than in men (17.6 vs. 12.2%), but these differences were not significant. The prevalence of diabetes varies considerably among Pacific Islands, ranging from almost undetectable in rural Papua New Guinea to ⬃40% in Nauru (1–3). The rates among Tongans observed in this study are among the highest in the region, excluding Nauru. Disturbingly, most of the diabetes (⬎80%) was previously undiagnosed. A number of studies have shown that ⬃50% of people with type 2 diabetes are undiagnosed, but the ratio of undiagnosed diabetes was higher in ethnic groups, being highest in MexicanAmericans (16). Among Mauritians, Dowse et al. (17) reported an overall ratio
of three undiagnosed for every two people with diagnosed diabetes, but they observed that this ratio was influenced by age, with the highest ratio of undiagnosed to diagnosed at 7.7 in women in the youngest age group. The high proportion of undiagnosed diabetes in Tonga probably reflects the lack of available routine medical services, the general lack of community awareness of diabetes, and the prevailing attitude of seeking medical advice only for advanced problems. A previous diabetes prevalence survey was conducted in 1973 on Tongatapu and Foa, an island of the Ha’apai group (18). The sampling methods for the survey population were similar to the current survey. A household census was performed 3 months before the survey, followed by random sampling within sex and 5-year age strata. However, the interpretation of the OGTT was different, with a fasting plasma glucose ⱖ6.7 mmol/l or a 2-h result ⱖ8.9 mmol/l being considered diagnostic of diabetes. The age of participants was similar to our study for Tongatapu (mean 40.8 years) but younger in Foa (mean 37.5 years). The previous study reported an age-adjusted prevalence for diabetes of 7.5%, 5% in men and 9.8% in women. No differences between the two locations were observed. Using the current diagnostic criteria, the previous study would have tended to overestimate the prevalence of diabetes. Nevertheless, it is clear that there has been an approximate doubling of diabetes over the past 25 years in Tonga. The lack of difference in diabetes prevalence between the more urbanized Tongatapu and the outer islands has been noted in both studies. This is likely to reflect that the degree of urbanization throughout Tonga is similar. Even in Tongatapu, many residents, including professionals, maintain a rural component to their lifestyle, and many continue to work small farms on weekends. The cardiovascular risk factor profile of people in Tongatapu and the islands was different— in the islands being overweight was common, but blood pressure was lower, lipid profile was better, and less people smoked. The classical risk factors for type 2 diabetes were also evident in Tonga, with significantly increased odds ratios for increasing age, weight expressed as BMI or WHR, hypertension, family history of diabetes, and microalbuminuria. However, DIABETES CARE, VOLUME 25, NUMBER 8, AUGUST 2002
Colagiuri and Associates
there was no association with any of the lipid measurements. These findings allow the generation of a risk profile for targeted screening for undiagnosed type 2 diabetes in Tongans. In addition to diabetes, IGT was also common, with an overall age standardized prevalence of 9.4%. However, the true rates of IGT may be nearly double this figure, because the survey methodology that only tested people with a fasting blood glucose level of ⬎5 mmol/l with an OGTT would have resulted in an underestimation of IGT. Indeed, 7.5% of people with a fasting blood glucose ⱕ5.0 mmol/l who had an OGTT had IGT, but none had diabetes. If this finding is applied to the whole population then the prevalence of IGT would be in the order of 17%. In contrast to undiagnosed diabetes and IGT, the prevalence of impaired fasting glycemia was only 1.6%. Comparison of the ADA and World Health Organization recommendations for diagnosing undiagnosed type 2 diabetes show that 75% of people would have been diagnosed based on the fasting plasma glucose alone, and another 16% would have been classified as having IFG. The overlap in diagnosis of previously undiagnosed type 2 diabetes observed in this study is similar to that reported by Shaw et al. (19). Therefore, in the Tongan population, use of the fasting plasma glucose alone would detect the majority of people with undiagnosed type 2 diabetes, and most of those who would not meet the diagnostic criteria for diabetes would be classified as IFG. Only 9% of people with undiagnosed diabetes based on the oral glucose tolerance would be classified as normal. In summary, the current prevalence of type 2 diabetes in Tonga is 15.1%, of which the majority (80%) is undiagnosed. The prevalence of diabetes has doubled since 1973. In addition, almost another estimated 20% of the population has lesser degrees of glucose intolerance. Much of the population is overweight and has other risk factors for cardiovascular
DIABETES CARE, VOLUME 25, NUMBER 8, AUGUST 2002
disease. These findings suggest that lifestyle-related noncommunicable disease will contribute the major source of burden of disease for the Tongan community unless interventions can be implemented to slow or reverse this trend. Acknowledgments — We are grateful to His Royal Highness, King Taufa’ahau Tupou IV, the Tongan Ministry of Health for its assistance with the survey, the survey teams from Tonga and Sydney, and the people of Tonga for their cooperation and participation. We are also indebted to AusAID for funding the survey. References 1. King H, Rewers M: Global estimates for prevalence of diabetes mellitus and impaired glucose tolerance in adults. Diabetes Care 16:157–177, 1993 2. Amos AF, McCarty DJ, Zimmet P: The rising global burden of diabetes and its complications: estimates and projections to the year 2010. Diabet Med (Suppl. 5):S1– S85, 1997 3. King H, Aubert RE, Herman WH: Global burden of diabetes, 1995–2025. Diabetes Care 21:1414 –1431, 1998 4. Zimmet P, King H, Taylor R, Raper LR, Balkau B, Borger J, Heriot W, Thoma K: The high prevalence of diabetes mellitus, impaired glucose tolerance and diabetic retinopathy in Nauru: the 1982 survey. Diabetes Res 1:13–18, 1984 5. Zimmet P, Faaiuso S, Ainuu J, Whitehouse S, Milne B, DeBoer W: The prevalence of diabetes in the rural and urban Polynesian population of Western Samoa. Diabetes 30:45–51, 1981 6. Zimmet P, Canteloube D, Genelle B, LeGonidec G, Couzigou P, Peghini M, Charpin M, Bennett P, Kuberski T, Kleiber N, Taylor R: The prevalence of diabetes mellitus and impaired glucose tolerance in Melanesians and PartPolynesians in rural New Caledonia and Ouvea (Loyalty Island). Diabetologia 23: 393–398, 1982 7. Zimmet P, Taylor R, Ram P, King H, Sloman G, Raper LR, Hunt D: Prevalence of diabetes mellitus and impaired glucose tolerance in biracial (Melanesian and Indian) population of Fiji: a rural-urban comparison. Am J Epidemiol 118:673–
688, 1983 8. King H, Taylor R, Zimmet P, Pargeter K, Raper LR, Beriki T, Tekanene J: Noninsulin-dependent diabetes (NIDDM) in a newly independent Pacific Nation: the Republic of Kiribati. Diabetes Care 7:409 – 415, 1984 9. King H, Taylor R, Koteka G, Nemaia H, Zimmet P, Bennett PH, Raper LR: Glucose tolerance in Polynesia: population-based surveys in Rarotonga and Niue. Med J Aust 145:505–510, 1986 10. Collins VR, Dowse GK, Toelupe PM, Imo TT, Aloaina FL, Spark RA, Zimmet PZ: Increasing prevalence of NIDDM in the Pacific Island population of Western Samoa over a 13-year period. Diabetes Care 17:288 –296, 1994 11. Bonapace U, Laureti L: Oceania index. In World Geographical Encyclopedia. Vol. 5. New York, McGraw-Hill, 1995 p. 34 –35 12. Department of Statistics Report. Nuku’alofa, Tonga, 1998 13. World Health Organization: Obesity-Preventing and Managing the Global Epidemic: Report of the WHO Consultation on Obesity. Geneva, World Health Org., 1997 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 20:1183–1197, 1997 15. World Health Organization: Definition, Diagnosis and Classification of Diabetes Mellitus and Its Complications. Part 1. Geneva, World Health Org., 1999 16. Harris MI: Undiagnosed NIDDM: Clinical and public health issues. Diabetes Care 16: 642– 653, 1993 17. Dowse GK, Gareeboo H, Zimmet PZ, Alberti KGMM, Tuomilehto J, Fareed D, Brissonnette LG, Finch CF: High prevalence of NIDDM and impaired glucose tolerance in Indian, Creole, and Chinese Mauritians. Diabetes 39:390 –396, 1990 18. Finau SA, Stanhope JM, Prior AM, Joseph JG: The Tonga cardiovascular and metabolic study: design, demographic aspects and medical findings. Community Health Studies 7:67–76, 1983 19. Shaw JE, de Courten M, Boyko EJ, Zimmet PZ: Impact of new diagnostic criteria for diabetes on different populations. Diabetes Care 22:762–766, 1999
1383