Age- and Sex-Specific Prevalence of Diabetes and Impaired Glucose Regulation in 11 Asian Cohorts Free

The study populations and the methods used to recruit the participants in the DECODA study have been previously reported (6,7). In the current data analysis, 11 population-based studies from four Asian countries were included (11–19). Among these, two studies were newly recruited, one from China, the Qingdao Diabetes Survey, and another from India, the National Urban Diabetes Study (NUDS) (19). Briefly, information on previous history of diabetes, data on glucose measurements at fasting and 2-h after a standard 75-g oral glucose tolerance test (OGTT) were sent to the Diabetes and Genetic Epidemiology Unit of the National Public Health Institute in Helsinki, Finland, where collaborative data analysis was carried out. The inclusion criteria for the current study are as follows: pure population-based studies (1); studies performed after 1980 (2); studies including both men and women (3); at least 20 years of age range (4); and a standard 2-h 75-g OGTT in the morning after an over night fast for at least 10 h (5), according to WHO recommendations (8,9).

A total of 22,171 subjects (9,806 men and 12,365 women) who had no previous history of diabetes and 2,164 subjects (1,045 men and 1,119 women) who had a previous history of diabetes from four countries in Asia met the inclusion criteria for the current data analysis. The age range was 30–89 years. The demographic characteristics and information on the OGTT is shown in Table 1 for each study cohort, and more detailed information is shown in the appendix. The proportion of female participants was slightly higher in 9 of the 11 studies. The participation rate for the 2-h OGTT varied from 72 to ∼100% in most of the studies. Plasma glucose was determined in 10 and capillary whole blood glucose in 1 of the studies. Before data analysis, the capillary whole blood glucose was converted to the plasma glucose according to the conversion factors described before (20).

Subjects who reported a previous history of diabetes were classified as previously diagnosed diabetic. Those who were not diagnosed as diabetic before screening were classified according to the 1999 WHO recommendations for the diagnosis of diabetes (5). The classification of undiagnosed diabetes, IGT, and normal glucose tolerance was made according to 2-h plasma glucose (2hPG) concentrations ≥11.1, 7.8–11.0, and <7.8 mmol/l, respectively. Fasting plasma glucose (FPG) concentrations ≥7.0, 6.1–6.9, and <6.1 mmol/l, respectively, classified subjects into diabetes, impaired fasting glucose (IFG), and normal fasting glucose. Subjects with IGT and/or IFG were defined as having impaired glucose regulation according to the WHO 1999 recommendations (5).

Mean fasting and 2hPG concentrations were calculated for the Chinese, Japanese, and Indian subjects, excluding patients with previously diagnosed diabetes (PDD). The cohorts with the same ethnic origin were combined as one population. Age- and sex-specific prevalences of diabetes were calculated for six 10-year age intervals, from 30 to 89 years, separately for people with and without a previous history of diabetes. The prevalence of PDD was calculated by dividing the number of diagnosed cases by the total number of subjects who had responded to the questions on the previous history of diabetes. Because there were some nonparticipants in the OGTTs, the prevalence of undiagnosed diabetes [(undiagnosed cases identified by the OGTTs/the number of individuals who attended the OGTT) × (1 − PDD)]. The same rule was applied for calculating the prevalence of IGR, IGT, and IFG. Because the prevalence pattern for Chinese and Japanese subjects was similar and differed markedly from that for Indian subjects, the data for Chinese and Japanese subjects were combined in the pooled analysis; similarly, the data for all Indian cohorts were combined. We further divided the Chinese and Japanese populations into urban and rural populations in order to make comparisons with the urban Indian population. A χ² test was used to measure the differences in the prevalence between men and women and between different age-groups.

Both FPG and 2hPG concentrations increased with age and reached a peak at 60–69 years of age then started to decline in Indian subjects and continued to increase after 70 years of age in Chinese and Japanese subjects (Fig. 1). At each age-group, the mean 2hPG was significantly higher for Indians than for Chinese and Japanese subjects (all P < 0.05), this was also true for FPG at most of the age-groups (Fig. 1). The increase is steeper for 2hPG than for FPG. The mean FPG and 2hPG concentrations did not differ between Chinese and Japanese subjects, except at 40–49 years of age where the glucose values were higher in Japanese than in Chinese subjects. There was no significant difference in mean glucose levels between men and women.

The prevalence of diabetes varied markedly among Asian populations studied. It increased with age up to 70–89 years of age in Chinese and Japanese subjects but up to 60–69 years of age and then declined thereafter in Indian subjects (Table 2 and Fig. 2). Age- and sex-specific prevalence, especially the peak prevalence of diabetes, was higher in cohorts of India and Singapore than in most of the Chinese and Japanese cohorts. In Chinese and Japanese subjects, the prevalence was <10% at 30–49 years of age; the peak prevalence was <20% in most of the cohorts, and none of the cohorts had a prevalence exceeding 30%. In contrast, in India and Singapore, the prevalence was >10% among those aged 40–49 years and over 30% among those aged 50–69 years for most of the cohorts. The urban Chinese and Japanese subjects had significantly higher prevalences of diabetes than their rural counterparts at 40–69 years of age for men and at 50–79 years of age for women (Fig. 2).

In all populations, the prevalence of isolated fasting hyperglycemia (FPG ≥7.0 mmol/l and 2hPG <11.1 mmol/l) did not increase with age (Fig. 2 and Table 2). The prevalence of isolated 2-h hyperglycemia (2hPG ≥11.1 mmol/l and FPG <7.0 mmol/l) tended to increase with age in Chinese and Japanese but not in Indian subjects.

The prevalence of IGR increased with age up to 70–89 years of age in most of the study cohorts (Table 3 and Fig. 3). The increase was graded with aging in Chinese, Japanese, and Singaporean populations but not in the Indian population, where the prevalence of IGR did not change much with age. The peak prevalences of IGR were not different among various populations, but the age-specific prevalence of IGR was higher in Indian than in Chinese and Japanese subjects at 30–49 years of age for both men and women. In the urban populations the prevalence was higher than in the rural population aged 40–69 years in men and 50–59 years in women (Fig. 3). In most of the study cohorts the age- and sex-specific prevalence of IGT was higher than the prevalence of IFG (Table 3 and Fig. 3). The concordance of the two was very poor.

The age-specific prevalence of diabetes did not differ between men and women except in a few age-groups where men had higher prevalences of diabetes than women (Table 2). The prevalence of IFG seemed to be higher in men than in women in Chinese and Japanese populations, whereas it was higher in women than in men in the Indian population (Table 3). The prevalence of IGT tended to be higher in women than in men in Chinese and Japanese subjects, but the difference was significant only at 40–49 years of age. There was no sex difference in the prevalence of IGT in Indian populations except at 30–39 years of age where women had higher prevalence of IGT than men.

The proportions of previously undiagnosed diabetes varied with age, with the highest in the youngest age-group and the lowest in the elderly. The declining trend with age was similar in Chinese, Japanese, and Indian subjects, with only one exception in the very old Indian population where the proportion increased again at 80–89 years of age. The proportion of undiagnosed diabetes was slightly higher in women than in men at younger age-groups. They were, in Chinese and Japanese men (women) combined, 0.77 (0.89), 0.59 (0.61), 0.58 (0.65), 0.55 (0.50), 0.40 (0.50), and 0.40 (0.35) and in all Indian men (women) combined 0.65 (0.71), 0.47 (0.49), 0.44 (0.43), 0.40 (0.40), 0.39 (0.44), and 0.60 (0.60), respectively, at 30–39, 40–49, 50–59, 60–69, 70–79, and 80–89 years of age.

We found that the prevalence of diabetes increased with age in all of our studies, but the peak prevalence of diabetes differed between different ethnic groups. The prevalence of diabetes reached its peak in the oldest age-group (>70 years) in Chinese and Japanese cohorts and at 60–69 years of age, followed by a decline at 70–79 years of age in Indian cohorts. Among people aged 30–79 years, the 10-year age-specific prevalence of diabetes was higher in Indian than in Chinese and Japanese subjects of the same age or even older. A younger age at onset of the disease has been previously reported in Asian Indian subjects (15,19). Asian Indian subjects have been identified as the ethnic group with one of the highest prevalences of diabetes (3,21). Also, the familial aggregation of type 2 diabetes is high among Indians, suggesting a strong genetic component for risk of the disease (22,23). Studies in Singapore (14) revealed that Asian Indian subjects were more centrally obese and insulin resistant and had low HDL and higher incidence of myocardial infarction than Chinese and Malay subjects living in Singapore. It is possible that the earlier onset of diabetes with a high prevalence and a high incidence of myocardial infarction resulted in a greater premature mortality in diabetic individuals, and that this contributed to the decline in the prevalence of diabetes after 70 years of age in Indian subjects.

In addition, we found that IGT was more prevalent than IFG in almost all age-groups in Asian subjects. The prevalence of IFG did not increase with age, but IGT increased more in the elderly, which is consistent with data in European populations (20). It is clear that a 2-h OGTT is an important part of a diabetes and IGR detection program in Asian populations. Without an OGTT the majority of the people with hyperglycemia would remain unidentified, and without intervention, this may prevent worsening from IGT to diabetes, as demonstrated by the Da Qing study (24).

Compared with the European populations included in the DECODE Study (20), the age-specific prevalence of diabetes in Urban Chinese and Japanese subjects was slightly higher than that in European subjects at 30–69 years of age, which is all lower than that in Indian subjects. The peak prevalence in the elderly population, however, was higher only in a few European populations, such as in Maltese, Finnish women in Oulu, and Spanish women living in the Canary Islands, than in Indian subjects. The age at which the peak prevalence of diabetes was reached was similar for European, Chinese, and Japanese subjects (over 70 years), whereas Indian subjects had their highest prevalence at 60–69 years of age. The prevalence of IGR increased gradually with age in Chinese and Japanese subjects, which was similar to that of European subjects but different from that of Indian subjects. In Indian subjects, the prevalence of IGR reached a high level already at 30 years of age and did not increase further with age. The difference in the prevalence of diabetes and IGR in different ethnic groups may not be completely explained by living environments and geographic locations, suggesting that genetic background play an important role.

The ratio of IGT to diabetes has been reported to decrease as prevalence of diabetes increases (3) and may have some predictive value in determining the stage of a glucose intolerance epidemic within a population (25). When the ratio is high but the prevalence of diabetes is low, the early stage of a diabetes epidemic may occur (3). The age- and sex-specific ratios of IGR to diabetes according to the newly revised diagnostic criteria for diabetes (5) were calculated for Asian and European populations (see data supplement at http://care.diabetesjournals.org). The ratio of IGR to diabetes seemed to decline when the prevalence of diabetes increased in both Asian and European populations, but the trend was not clear. It has been estimated that there will be a 57% increase in the prevalence of diabetes in Asia and a 24% increase in Europe from the year 2000 to the year 2010 (26). According to this, a higher ratio of IGR to diabetes will be expected in Asia than in Europe given the same age and same prevalence of diabetes. This was, however, not observed. Despite the low ratio of IGT to diabetes in India, a recent study showed that both the prevalence of diabetes and IGT are still increasing nation-wide in India (19).

In conclusion, the age- and sex-specific prevalence of diabetes was higher in Indian than in Chinese and Japanese subjects, but the prevalence of IGR was not, except in very young populations. The prevalence of diabetes increased with age, reaching its peak at 70–89 years of age in Chinese and Japanese subjects, and at 60–69 years of age followed by a decline at 70 years of age in Indian subjects. IGT was more prevalent than IFG in all age-groups in Asians. We strongly recommended retaining the 2-h OGTT when determining impaired glucose regulation in Asian populations.

The DECODA (Diabetes Epidemiology: Collaborative Analysis Of Diagnostic Criteria in Asia) study was initiated in 1998 and carried out under the supervision of the International Diabetes Epidemiology Group. (Jaakko Tuomilehto, National Public Health Institute, Helsinki, Finland; Knut Borch-Johnsen, Steno Diabetes Center, Gentofte, Denmark; Naoko Tajima, Jikei University School of Medicine, Tokyo, Japan; and Clive S. Cockram, Department of Medicine, the Chinese University of Hong Kong, Prince of Wales Hospital, Hong Kong, China).

Qing Qiao, Jaakko Tuomilehto, Department of Epidemiology and Health Promotion, National Public Health Institute, Helsinki, Finland; Tomoko Nakagami, Knut Borch-Johnsen, Steno Diabetes Center, Gentofte, Denmark; and Takeshi Kuzuya, JA Shioya General Hospital, Tochigi, Japan.

Qing Qiao, Gang Hu, Department of Epidemiology and Health Promotion, National Public Health Institute, Helsinki, Finland.

Qing Qiao,¹ Gang Hu,¹ Jaakko Tuomilehto,¹ Tomoko Nakagami,² Beverley Balkau,³ Knut Borch-Johnsen,⁴ A. Ramachandran,^5,15 V. Mohan,⁶ S. Ramnathan Iyer,⁷ Makoto Tominaga,⁸ Yutaka Kiyohara,⁹ Isao Kato,⁹ Ken Okubo,⁹ Masaki Nagai,¹⁰ Satomi Shibazaki,¹⁰ Ze Yang,^11,12 Zhifu Tong,¹¹ Qinying Fan,¹¹ Binzhi Wang,¹² Suok Kai Chew,¹³ Bee Yian Tan,¹³ Derrick Heng,¹³ Shanta Emmanuel,¹³ Naoko Tajima,¹⁴ Yasuhiko Iwamoto,² C. Snehalatha,¹⁵ V. Vijay,¹⁵ A. Kapur,¹⁵ Yanhu Dong,¹⁶ Hairong Nan,¹⁶ Weiguo Gao,¹⁶ Haiyan Shi,¹⁶ Fangming Fu¹⁶.

From the ¹Diabetes and Genetic Epidemiology Unit, Department of Epidemiology and Health Promotion, National Public Health Institute, Helsinki, Finland; the ²Diabetes Center, Tokyo Women’s Medical University, Tokyo, Japan; ³INSERM U258, Paris, France, the ⁴Steno Diabetes Center, Gentofte, Denmark; the ⁵Chennai 94 Study: Diabetes Research Center, Chennai, India; the ⁶Chennai 97 Study: Madras Diabetes Research foundation, Chennai, India; the ⁷Dombivli Study: Ambika Clinic and Research Center, Dombivli (East), District Thane, Maharashtra, India; ⁸Funagata Diabetes Studies: Department of Laboratory Medicine, Yamagata University, Yamagata, Japan; the ⁹Hisayama Study: the Department of Medicine and Clinical Science, Graduate School of Medical Science, Kyushu University, Fukuoka, Japan; ¹⁰Ojika Surveys: Department of Public Health, Saitama Medical School, Saitama, Japan; the ¹¹Beijing Study: Institute of Geriatrics and Beijing Hospital and Chinese-Japanese Friendship Hospital, Ministry of Health, Beijing, China; the ¹²Shunyi Study: Shunyi Country People’s Hospital, Shunyi, China; the ¹³National Health Survey: Epidemiology and Disease Control, Ministry of Health, Singapore; the ¹⁴Jikei University School of Medicine, Tokyo, Japan; ¹⁵The National Urban Diabetes Survey (NUDS), India; and the ¹⁶Qingdao Endocrinology and Diabetes Institute, Hiserve Medical Group, Qingdao, China.

The DECODA Study was initially funded by Novo Nordisk, Bagsvaerd, Denmark. This analysis has been carried out with the help of grants from Novartis Pharma, Basel, Switzerland, and from the Finnish Academy (Grants 46558, 76502, and 77618). The Japan Diabetes Foundation and Japan Arteriosclerosis Prevention Fund (Japanese Ministry of Health, Welfare and Labor) and Shiseikai Overseas Research Foundation (Tokyo Women’s Medical University) supported a fellowship for Tomoko Nakagami.