Prevalence of Diabetes and Impaired Fasting Glucose in Korea: Korean National Health and Nutrition Survey 2001

The Korean National Health and Nutrition Survey (KNHNS) was performed by the Korea Institute for Health and Social Affairs for the Korean Ministry of Health and Welfare. The 2001 KNHNS was a cross-sectional and nationally representative survey conducted from 1 November through 30 December 2001 and consisted of four components: a Health Interview Survey, a Health Behavior Survey, a Health Examination Survey, and a Nutrition Survey. The target population for this survey was civilian noninstitutionalized individuals aged >1 year (for the Health Interview and Nutrition Surveys) or aged >10 years (for the Health Behavior and Health Examination Surveys). A stratified multistage probability sampling design was used, and selections were made from sampling units based on geographical area, sex, and age using household registries. There were 246,097 primary sampling units, each of which contained about 60 households. Two hundred sampling frames (12,180 households) from primary sampling units were randomly sampled throughout South Korea, and 37,769 individuals from these sampling frames were included in the Health Interview Survey. The survey was completed by 9,770 of 12,642 individuals who participated in the Health Examination Study, which gave a participation rate of 77.3%. However, information on the total number of dropouts, including refusals and noncontacted subjects, was not obtained, and thus the response rate could not be determined. We used the data of 6,601 subjects aged >20 years for this study. Each respondent was assigned a weight, based on geographic and demographic characteristics, to allow findings to be extrapolated for the entire Korean population. Nurses were trained to perform anthropometric measurements, serum collection, and blood pressure measurement and to mange questionnaires. The questionnaires included questions concerning the demographic, socioeconomic, dietary, and medical history details for each respondent.

Body weight and height were measured with subjects wearing light clothing without shoes. BMI was calculated as weight in kilograms divided by the square of height in meters. Waist circumference was measured from the narrowest point between the lower borders of the rib cage and the iliac crest. Blood pressure was measured in the sitting position after a 10-min rest period. Two systolic and diastolic blood pressure readings were recorded with a 5-min interval and averaged for analysis.

Fasting blood sample were taken in the morning after at least an 8-h fast. Blood samples were centrifuged, refrigerated at the examination site, and transferred in iceboxes to a central laboratory in Seoul on the day taken. Plasma glucose, total cholesterol, triglyceride, and HDL cholesterol levels were measured using an autoanalyzer (Hitachi 747 autoanalyzer, Tokyo, Japan). LDL cholesterol was calculated using the Friedewald equation for those with serum triglyceride levels ≤400 mg/dl (11).

Diabetes was defined as a fasting plasma glucose value ≥7.0 mmol/l (126 mg/dl) or a previous diagnosis of diabetes. Diabetes was further subclassified as “known diabetes” if diabetes had been diagnosed previously by a medical practitioner or as “newly diagnosed diabetes” if diabetes was first diagnosed by this study. IFG was defined as a fasting plasma glucose value of 5.6–6.9 mmol/l (100–125 mg/dl) in the absence of a previous diagnosis of diabetes, in accord with the new American Diabetes Association (ADA) criteria (12).

Self-reported alcohol intake, smoking, and physical exercise were estimated from the questionnaire. Alcohol consumption was estimated from the usual intake amount and frequency of alcoholic beverage intake as follows: none, ≥1 cup once a week, ≥5 cups in men or ≥4 cups in women once or twice a week, or ≥5 cups in men or ≥4 cups in women three times a week or more (13). In terms of smoking, individuals were classified as nonsmokers, ex-smokers, or current smokers. Physical exercise was defined according to estimated energy consumption per week: none, <7.5 kcal · kg⁻¹ · week⁻¹, 7.5–15 kcal · kg⁻¹ · week⁻¹, and >15 kcal · kg⁻¹ · week⁻¹ (13). Using the definition applied by the Korean government, small towns and villages with a population of ≤50,000 were designated as rural areas (14). A family history of diabetes was considered positive if at least one parent or one sibling had diabetes.

The demographic characteristics of study subjects are presented using means ± SD or as numbers and percentages, as appropriate, for normal glucose, IFG, and diabetic subject groups. The age-adjusted prevalence of cardiovascular risk factors for each fasting glucose group were obtained using the direct adjustment method of standardization. National census data from the Korea National Statistical Office in 2001 was used to define the standard Korean population. Survey weights were taken into account to calculate prevalence standard errors. The linear effect of polynomial contrasts of the prevalence across glycemia categories was also examined. Polychotomous logistic regression analysis using a multinomial logit model was used to identify risk factors of diabetes or IFG. The independent variables used in this multivariable analysis were as follows: age, sex, BMI, systolic blood pressure, diastolic blood pressure, total cholesterol, triglyceride, HDL cholesterol, residential area, education level, monthly income, drinking status, smoking status, exercise status, and a family history of diabetes. Linear trend tests of odds ratios (ORs) according to each of the categorical variables in the model were also determined. All statistical analyses were performed using SUDAAN release 8.0 (Research Triangle Institute, Research Triangle Park, NC) to reflect the characteristics of the study’s multistage sampling design. Reported P values were two-tailed, and P < 0.05 was considered to be statistically significant.

A total of 6,601 participants >20 years of age were included in this analysis. After subjects without fasting glucose information (n = 128), those who fasted for too short a time (n = 401), and those without fasting time data (n = 236) were excluded, 5,836 subjects remained, and of these, 8 subjects met two exclusion criteria, and thus 5,844 subjects (2,513 men and 3,331 women) were finally included in the analysis. The age-adjusted prevalence of diabetes was 7.6% (men, 8.1%; women, 7.5%). The age-standardized prevalences of self-reported diabetes were similar in men and women (4.4 and 4.5%, respectively). We found that 3.3% of subjects had previously undiagnosed diabetes (3.7% in men and 3.0% in women). In addition, age-adjusted prevalence of IFG was 23.9% (26.3% in men and 22.6% in women) in the study population according to the new ADA criteria. In contrast, when the previous IFG criterion (fasting plasma glucose 6.1–6.9 mmol/l) was used, this age-adjusted prevalence of IFG fell to 7.0% (7.9% in men and 6.5% in women). Based on prevalence estimates and national census data in 2001 from the Korea National Statistical Office, the numbers of Korean adults with diabetes or IFG were estimated to be 2.6 and 8.1 million, respectively. Moreover, it was estimated that diabetes was undiagnosed in 43.0% of those individuals.

The age-specific prevalence of diabetes increased with age and reached its peak in the oldest female age-group. However, there was a peak of age-adjusted prevalence of diabetes in men 60–69 years of age, and, after that, it decreased slightly in those ≥70 years (Fig. 1). Moreover, in men, diabetes prevalence increased steadily, whereas in women it surged abruptly at 50 years of age. Unlike the prevalence of diabetes, the prevalence of IFG did not increase significantly with age.

Table 1 presents the demographic and biological characteristics of the study subjects classified according to fasting glucose level. In Table 2, the age-adjusted prevalence of cardiovascular risk factors according to glycemia categories are presented for men and women. Compared with subjects with normal plasma glucose, men with IFG or diabetic men showed higher blood pressure, BMI, waist circumference, and triglyceride levels. In women with IFG or diabetic women, trends of increasing blood pressure, BMI, waist circumference, total cholesterol level, triglyceride level, and decreased HDL cholesterol level were observed (Table 2).

ORs and their 95% CIs for diabetes and IFG compared with the normal glucose group are shown in Table 3. Using polychotomous multiple logistic regression analysis, diabetes was found to be associated with age, BMI, blood pressure, triglyceride, HDL cholesterol, education level, alcohol consumption, exercise, and a family history of diabetes, whereas IFG was associated with BMI, systolic blood pressure, total cholesterol, HDL cholesterol, residential area, education level, alcohol consumption, and a family history of diabetes. In terms of alcohol consumption, only high-risk drinking was found to be significantly associated with both diabetes and IFG.

The present study provides the first representative, population-based estimates of the prevalence of diabetes and IFG in Korean adults. The KNHNS 2001 results indicate that the age-adjusted prevalence of diabetes in the Korean population is 7.6%, which includes 4.4% previously diagnosed and 3.3% newly diagnosed. These prevalences of diabetes are comparable to those of Western countries such as the U.S. (7) and Australia (15). Recent Asian studies have shown diabetes prevalences of 5.5% in China (16) and 9.6% in Thailand (17). In a study conducted in rural Japan, the prevalences of diabetes were found to be 9.1% for men and 10.8% for women (18). Several previous studies have yielded prevalence estimates for diabetes in Korea. In 1993, Korean diabetes prevalence was reported to be 7.2% in a community-based cross-sectional study (8). Another Korean study reported a prevalence of diabetes in a rural area of 7.9% according to ADA criteria (19). However, these studies were based on rural populations.

The KNHNS was designed to provide a reliable nationwide estimate of health-, nutrition-, and chronic disease-related basic statistics and was performed on a large representative sample of the Korean population. Moreover, reliable sampling designs and standardized data collection protocols were used to assure accurate estimates of diabetes in the Korean population.

In the present study, the prevalence of diabetes increased with age and peaked in the oldest age-group; however, IFG prevalence did not show this trend. These results are comparable to those previously reported in China and Japan (6).

In this study, the prevalence of IFG was determined to be 23.9% (26.3% in men and 22.6% in women) using the new ADA criteria (fasting plasma glucose 5.6–6.9 mmol/l). This is considerably higher than expected in both men and women. With the previous criteria (fasting plasma glucose 6.1–6.9 mmol/l), this prevalence was 7.0% (7.9% in men and 6.5% in women), which is similar to or higher than those reported in other Asian countries (16,17) and might predict a future diabetes epidemic in the Korean population. Thus, the findings of the present study have important public health implications. Diabetes gives rise to micro- and macrovascular complications and results in an enormous economic burden. Patients with IFG are now referred to as having “pre-diabetes,” which reflects a high risk of diabetes development. IFG is not a clinical entity but rather a risk factor for future diabetes and cardiovascular disease (12). Furthermore, IFG is known to be associated with the metabolic syndrome, which includes components of obesity, dyslipidemia, and hypertension. Therefore, public attention to the high prevalence of IFG and diabetes is requested.

This study endorses the fact that established risk factors, such as aging, hypertension, dyslipidemia, and a family history of diabetes, are associated with diabetes in the Korean population and concurs with the results of studies in Western populations in this respect (20,21). Moreover, when only patients with newly diagnosed diabetes are considered (i.e., by excluding subjects with known diabetes from the analysis), its association with normal subjects became insignificant for age, triglyceride level, and a family history of diabetes, whereas the significances of other variables remain unchanged (data not provided). Previous studies in the U.S. and Europe have reported an inverse correlation between alcohol intake and type 2 diabetes (22) or have suggested no significant association with diabetes (23). However, a recent study revealed that moderate to high alcohol consumption is positively associated with the incidence of diabetes in lean Japanese men (24). In the present study, only heavy drinkers were found to be at risk of diabetes and IFG.

Several limitations of this study need to be considered. First, no oral glucose tolerance tests (OGTTs) were conducted, which indicates that the prevalence of diabetes could have been underestimated. In our previous study of elderly Koreans, the prevalence of newly diagnosed diabetes was higher according to World Health Organization (WHO) criteria using OGTTs than by ADA criteria using fasting glucose alone (9). In addition, several studies have indicated that IFG might not be as sensitive a predictor of diabetes as impaired glucose tolerance (25,26). OGTTs are capable of identifying individuals at increased with risk of cardiovascular complications who would not be detected by fasting glucose testing alone. Furthermore, the ADA and WHO criteria might identify different groups of individuals. In the Diabetes Epidemiology: Collaborative Analysis of Diagnostic Criteria in Europe (DECODE) study, ADA criteria were found to be more likely to identify obese individuals ≤65 years old, whereas WHO criteria showed a tendency to diagnose diabetes in lean individuals (27). Moreover, OGTT is less reproducible than fasting plasma glucose testing (28). In an epidemiological study, use of a fasting glucose measurement is recommended because of its convenience and reproducibility (12). Second, the present study was limited by its cross-sectional nature, which prevents our defining causal relationships. Third, this survey does not provide information on subject refusals and/or noncontacted subjects, and thus, even though the participant rate was 77.3%, the response rate could not be calculated. Nevertheless, this study does provide the most reliable up-to-date information on the prevalences of diabetes and IFG and on their relationships with cardiovascular risk factors in the Korean population.

Summarizing, the present study shows that diabetes and IFG are common and that they are associated with cardiovascular risk factors in Korean adults. Without vigorous efforts to prevent and treat diabetes, the socioeconomic burden is predicted to be enormous. Prompt public action is requested to control diabetes in Korea.

We thank the members of the Korea Institute for Health and Social Affairs who conducted the national survey and everyone who contributed to this project.