Prevalence and Ethnic Pattern of Diabetes and Prediabetes in China in 2013

Key Points

Question  What is the prevalence of diabetes and prediabetes in China?

Findings  In this nationally representative cross-sectional survey conducted in 2013 in mainland China with 170 287 participants, the estimated overall prevalence of total diabetes was 10.9%; of prediabetes, 35.7%. Tibetan and Muslim Chinese individuals had significantly lower prevalence of diabetes than Han participants.

Meaning  The estimated prevalence of diabetes and prediabetes in China differs by ethnic group, and the estimated overall high prevalence may be associated with different methods of measuring hemoglobin A_1c.

Importance  Previous studies have shown increasing prevalence of diabetes in China, which now has the world’s largest diabetes epidemic.

Objectives  To estimate the recent prevalence and to investigate the ethnic variation of diabetes and prediabetes in the Chinese adult population.

Design, Setting, and Participants  A nationally representative cross-sectional survey in 2013 in mainland China, which consisted of 170 287 participants.

Exposures  Fasting plasma glucose and hemoglobin A_1c levels were measured for all participants. A 2-hour oral glucose tolerance test was conducted for all participants without diagnosed diabetes.

Main Outcomes and Measures  Primary outcomes were total diabetes and prediabetes defined according to the 2010 American Diabetes Association criteria. Awareness and treatment were also evaluated. Hemoglobin A_1c concentration of less than 7.0% among treated diabetes patients was considered adequate glycemic control. Minority ethnic groups in China with at least 1000 participants (Tibetan, Zhuang, Manchu, Uyghur, and Muslim) were compared with Han participants.

Results  Among the Chinese adult population, the estimated standardized prevalence of total diagnosed and undiagnosed diabetes was 10.9% (95% CI, 10.4%-11.5%); that of diagnosed diabetes, 4.0% (95% CI, 3.6%-4.3%); and that of prediabetes, 35.7% (95% CI, 34.1%-37.4%). Among persons with diabetes, 36.5% (95% CI, 34.3%-38.6%) were aware of their diagnosis and 32.2% (95% CI, 30.1%-34.2%) were treated; 49.2% (95% CI, 46.9%-51.5%) of patients treated had adequate glycemic control. Tibetan and Muslim Chinese had significantly lower crude prevalence of diabetes than Han participants (14.7% [95% CI, 14.6%-14.9%] for Han, 4.3% [95% CI, 3.5%-5.0%] for Tibetan, and 10.6% [95% CI, 9.3%-11.9%] for Muslim; P < .001 for Tibetan and Muslim compared with Han). In the multivariable logistic models, the adjusted odds ratios compared with Han participants were 0.42 (95% CI, 0.35-0.50) for diabetes and 0.77 (95% CI, 0.71-0.84) for prediabetes for Tibetan Chinese and 0.73 (95% CI, 0.63-0.85) for diabetes and 0.78 (95% CI, 0.71-0.86) for prediabetes in Muslim Chinese.

Conclusions and Relevance  Among adults in China, the estimated overall prevalence of diabetes was 10.9%, and that for prediabetes was 35.7%. Differences from previous estimates for 2010 may be due to an alternate method of measuring hemoglobin A_1c.

China has the world’s largest diabetes epidemic, which continues to increase.¹ The prevalence of diabetes in China was reported to be 0.67% in 1980 and 11.6% in the latest published nationwide estimate in 2010.^2-5 Moreover, according to the 2010 survey, the prevalence of prediabetes was 50.1%, implying that approximately 500 million Chinese adults may have had prediabetes.⁵ However, this high estimate for prediabetes raised concern over the possibility of an overestimation.^6,7 As a disorder of glucose metabolism, diabetes mellitus affects multiple organ systems and is associated with a variety of vascular and several nonvascular complications.^8-10 Ongoing reliable estimations are needed to plan effective national prevention and treatment programs for diabetes management.

In addition, although a few epidemiologic studies were available to estimate the prevalence of diabetes among different ethnic groups in China, the data were from different studies conducted at different times.^11-13 Direct comparison under a consistent survey design is therefore of importance for policy making for diabetes management in Chinese minorities.¹⁴

This study was conducted to provide more recent estimates of the prevalence of diabetes and prediabetes in China and to investigate their ethnic pattern, using a nationally representative survey conducted in 2013.

The China Chronic Disease and Risk Factors Surveillance study is a nationwide cross-sectional study conducted every 3 years. The study was designed to measure the epidemiology of chronic disease and associated risk factors by selecting a nationally representative sample of the general population.^5,15,16 In the 2013 survey, an integrated national mortality surveillance system was used to provide a nationally and provincially representative survey of the country.¹⁷ This new system increased the number of surveillance points from 161 to 605 to cover major geographic areas of all 31 provinces, autonomous regions, and municipalities in mainland China, and enlarged the surveillance population covered from 6% to 24% of the Chinese population. The first level of sampling was stratified by 31 provinces. Within each province, 8 strata were then generated according to the overall population size (high/low), proportion of urban population (high/low), and mortality rate (high/low). A total of 298 surveillance points were used. The complex multistage sampling design, stringent quality assurance, and quality control program were previously reported.^5,16 Participants (179 347) were enrolled in the survey from 1176 rural townships or urban subdistricts. The overall response rate with the standard definition by the American Association for Public Opinion Research was 93.4% (the cooperation rate was 96.2% and the refusal rate was 2.7%) (eTable 1 in the Supplement).¹⁸ Overall, 6.25% of the sampled families could not be accessed during 3 attempts made on 3 days. These households were replaced by another household with a similar family structure in the same village or residential area (eTable 2 in Supplement). Only persons who had been living in their current residence for at least 6 months within 1 year before the survey and were aged 18 years or older were eligible for the survey. Exclusion criteria were communal residences (eg, university dormitory, military unit, nursing home); patients with severe diseases or having cognitive, language, or mental disorders who could not participate in the interview; and pregnancy. Additionally, we restricted the analytic sample to participants with complete information on major risk factors (ie, age, sex, location, economic development level, body mass index [BMI], and history of diabetes or glycemia markers). This analysis included 170 287 participants (94.9% of those enrolled); 9060 participants were excluded because of missing information on major risk factors.

The study protocol was approved by the ethical review committee of the Chinese Center for Disease Control and Prevention and other participating institutes. Written informed consent was obtained from all study participants.

A comprehensive questionnaire including information on demographic characteristics, medical history, and lifestyle risk factors was administered by trained interviewers. Staff at local health stations or community clinics in the participants’ residential area were trained according to a standard protocol. Information on the ethnic groups of participants was self-reported but cross-checked with participants’ national identification. There are 56 ethnic groups in China. The 10 largest ones were listed in the questionnaire and others were collected by open-ended questions. Analyses by ethnic groups were restricted to those with at least 1000 participants, resulting in inclusion of Tibetan (Zang), Zhuang, Manchu (Man), Uyghur (Wei), and Muslim (Hui) ethnic groups. Each surveillance point was categorized into underdeveloped, intermediately developed, or developed according to the region’s gross domestic product per capita in 2013. Blood samples were collected for all participants after an overnight fast of at least 10 hours. Serum samples and measurements of fasting plasma glucose level and 2-hour plasma glucose level after 75 g glucose were managed under the same procedure used in the 2010 survey, as reported previously.⁵ In contrast, to overcome limitations in the calculated level of venous hemoglobin A_1c (HbA_1c) used previously, HbA_1c was directly measured from venous blood samples in the 2013 survey in a single laboratory using quantitative high-performance liquid chromatography and the boronate affinity method (Bio-Rad D-10 Hemoglobin Analyzer). Venous blood samples were stored at –80°C until HbA_1c was measured within 1 month, which is within the range of stability according to the manufacturer’s instructions. Detailed information related to the HbA_1c assay method and quality control is described in the eAppendix in the Supplement.

Diagnosed diabetes was defined as a self-reported diagnosis that was determined previously by a health care professional. Undiagnosed diabetes was defined according to the American Diabetes Association 2010 criteria, which included fasting plasma glucose level of 126 mg/dL or greater, 2-hour plasma glucose level of 200 mg/dL or greater (after 75 g glucose), or HbA_1c level of 6.5% or higher. Total diabetes was the sum of the number of patients who received a diagnosis of diabetes and the number of patients with undiagnosed diabetes. Prediabetes was defined as any participants who did not have diabetes but who had an HbA_1c level of 5.7% to 6.4%, fasting plasma glucose level of 100 to 125 mg/dL, or 2-hour plasma glucose level of 140 to 199 mg/dL. Awareness was defined as the proportion of individuals with physician-diagnosed diabetes among all patients with diabetes. Treatment was defined as the proportion of individuals receiving diabetes medications among all patients with diabetes. Control was defined as the proportion of individuals with an HbA_1c concentration of less than 7.0% among patients with diabetes who were taking medication.

Standardized prevalence was estimated in the overall population and in subgroups of sex, age, location (urban/rural), stages of economic development, and categories of BMI in the Chinese adult population. The subgroup variables were selected before the analyses were conducted. Standardized prevalence was calculated unless stated otherwise, using the weight coefficients to represent the overall Chinese adult population aged 18 years or older. Weight coefficients accommodated the sampling scheme for unequal probabilities of sample selection, as well as the poststratification weights, which harmonized the sample structure of the survey with that of the 2010 Chinese population census. For the analyses of ethnic variation, unweighted estimations are presented because ethnicity was not considered in the multistage sampling design. The χ² test was used to compare crude prevalences. A number of participants (9060) were excluded from the analyses because of missing important information. Multivariable logistic regression was used to examine the association of ethnic groups with the odds of diabetes and prediabetes. Five models with progressively increased adjustment of risk factors were applied. All P values were 2-tailed and were adjusted for multiple testing, and 95% CI was used. P < .05 was the threshold for statistical significance. All statistical analyses were conducted with SAS version 9.4 and Stata version 14.1.

Data were available for 170 287 participants aged 18 years or older (Table 1). The basic characteristics of participants included and excluded in the analyses are presented in eTable 3 in the Supplement, which demonstrate similar age, sex, and BMI, with a higher prevalence of lower educational status and a higher activity level in the included group. The mean age at baseline was 43.5 years (SD 16.2), BMI was 24.0, and 57% were women. The mean fasting plasma glucose level was 100.5 mg/dL, and the 2-hour plasma glucose level was 114.2 mg/dL, similar to those in the 2010 China Chronic Disease and Risk Factors Surveillance study, whereas the mean HbA_1c levels decreased from 5.8% in 2010 to 5.38% in 2013 (P < .001). The overall standardized prevalence of total diabetes in Chinese adults was estimated as 10.9% (95% CI, 10.4%-11.5%), with 10.2% (95% CI, 9.7%-10.7%) in women and 11.7% (95% CI, 10.9%-12.4%) in men (Table 2). Four percent (95% CI, 3.6%-4.3%) of Chinese adults had previously received a diagnosis of diabetes; 6.9% (95% CI, 6.7%-7.2%) of the population had received a new diagnosis by glycemic biomarkers (ie, fasting plasma glucose, 2-hour plasma glucose, and HbA_1c levels). Hemoglobin A_1c concentration added an additional 0.5% to the total diabetes group beyond those previously receiving a diagnosis of diabetes or through traditional glycemic biomarkers. The prevalence of diabetes was higher in the older population (P < .001), men (P < .001), urban residents (P < .001), participants living in economically developed areas (P = .003), and overweight and obese participants (P < .001). Prevalence estimates were also higher in overweight and obese participants when Asian-specific BMI cutoffs were used (P < .001) (Table 2).

The estimated prevalence of prediabetes was 35.7% (95% CI, 34.1%-37.4%) in the overall population, 35.0% (95% CI, 33.4%-36.7%) in women and 36.4% (95% CI, 34.6%-38.2%) in men. The prevalence of prediabetes was higher in the older population (P < .001; men, P = .008) and overweight and obese participants (P < .001), particularly with Asian-specific cutoffs (P < .001). Prediabetes was more prevalent in rural residents than in urban residents (P = .02) (Table 2).

The overall proportion of patients who were aware of their diabetes condition was 36.5% (95% CI, 34.3%-38.6%) (Table 3). The proportion of women who were aware of their diabetes condition was 39.8% (95% CI, 37.5%-42.2%), and of men it was 33.5% (95% CI, 31.2%-35.9%); 32.2% (95% CI, 30.1%-34.2%) of all patients with diabetes were receiving antidiabetic medication. Among patients treated, 49.2% (95% CI, 46.9%-51.5%) had HbA_1c levels controlled to a concentration of less than 7.0%.

The proportion of patients who were aware of their diabetes and treated for it was higher in the older population (P < .001 for both), women (P < .001 for both), and urban residents (P < .001 for both). The proportion of patients who controlled their HbA_1c levels well was higher in younger individuals (P = .03) and in urban residents (P < .001), but broadly similar across sex, BMI categories, and economic development conditions. The characteristics of treated diabetes patients with and without adequate glycemic control were also compared (eTable 4 in the Supplement). Of patients without adequate glycemic control, a higher percentage received insulin treatment.

Compared with Han participants, whose crude diabetes prevalence was 14.7% (95% CI, 14.6%-14.9%), the crude prevalence of total diabetes was lower among major minority ethnic groups, except for Manchu participants (Table 4). Tibetan and Muslim adults had the lowest and second lowest prevalence of total diabetes (4.3% [95% CI, 3.5%-5.0%] vs 10.6% [95% CI, 9.3%-11.9%]; both P < .001 compared with Han participants).

For prediabetes, Tibetan and Muslim adults had a similar crude prevalence that was significantly lower than that of Han participants (31.3% [95% CI, 29.7%-32.9%] for Tibetan, 31.9% [95% CI, 29.9%-33.9%] for Muslim, and 38.8% [95% CI, 38.5%-39.0%] for Han; Tibetan and Muslim P < .001 compared with Han) (Table 5). Manchu participants had a significantly higher prevalence of prediabetes than did Chinese Han (43.4% [95% CI, 41.3%-45.5%]; P < .001).

In the multivariable logistic models for diabetes, the age- and sex-adjusted odds ratio for Tibetan Chinese was 0.34 (95% CI, 0.29-0.41); for Muslim Chinese, 0.76 (95% CI, 0.66-0.88) (Table 6). In the fully adjusted multivariable logistic models, the adjusted odds ratios compared with Han participants were 0.42 (95% CI, 0.35-0.50) for diabetes and 0.77 (95% CI, 0.71-0.84) for prediabetes for Tibetan Chinese and 0.73 (95% CI, 0.63-0.85) for diabetes and 0.78 (95% CI, 0.71-0.86) for prediabetes for Muslim Chinese.

In this large, nationally representative survey of Chinese adults in 2013, it was estimated that 10.9% of Chinese adults have diabetes. The study has yielded several findings about the prevalence and ethnic variation of diabetes among the major minorities in China.

The prevalence of diabetes, at 10.9%, was only slightly lower than the prevalence of total diabetes in the US population (12%-14%) in 2011-2012, although the mean BMI was still much lower than that of the US population (24.0 vs 28.7).¹⁹ The prevalence of diabetes was much higher in overweight BMI categories in this study than in the US population (15.4% vs 8%-9%). This was consistent with the finding that Asians may have a higher risk of developing diabetes at a given BMI. However, when the Asian-specific BMI cutoffs were applied, the prevalence of diabetes for overweight persons was more comparable with that of the US population (Table 2).

The updated prevalence of prediabetes in 2013 was significantly lower than that in 2010 (35.7% vs 50.1%). With approximately 1.09 billion adults in total in mainland China, it is projected that 388.1 million Chinese adults (200.4 million men and 187.7 million women) may have had prediabetes in 2013, considerably reduced from the projected number of 493.4 million individuals in 2010.⁵ Projections were calculated with the overall Chinese population aged 18 years and older in 2013 multiplied by the estimated prevalence based on the sampling weight. Sex-specific projections used sex-specific estimations of the prevalence. The 2010 and 2013 surveys were conducted under the same study protocol and followed identical, strict strategies for quality control within the serial cross-sectional China Chronic Disease and Risk Factors Surveillance study. The most likely reason for the discrepancy in prediabetes prevalence was a difference in HbA_1c measurement between the 2 surveys. Hemoglobin A_1c level was calculated in 2010 by converting capillary HbA_1c with a formula derived from an internal validation study, whereas HbA_1c was directly measured from venous blood samples stored at –80°C in the 2013 survey, which is a more reliable method. The estimated prevalence of prediabetes in the United States was 36.5% in 2011-2012,¹⁹ similar to that in the 2013 Chinese survey. Overall, 47% of the Chinese adult population was estimated to have either diabetes or prediabetes, slightly lower than the 49% to 52% estimate in the US population.

China has 56 ethnic groups, with Chinese Han the majority. There are substantial differences in genetic background, culture, socioeconomic levels, climate and geographic features of the residential area, lifestyle, and dietary pattern among certain ethnic groups. In the 2010 census, the populations of Tibetan, Zhuang, Manchu, Uyghur, and Muslim were 6.3, 16.9, 10.4, 10.1, and 10.6 million, respectively. Variants of 2 genes involved in oxygen processing were found in most Tibetan highlanders, who thrive at altitudes greater than 4400 m.²⁰ Significant variation in dietary patterns among different ethnic groups also has been reported.²¹ The relatively large sample size of the survey provided enough participants enrolled in major Chinese minority groups. Compared with Han participants, Tibetan adults had a significantly lower prevalence of both total diabetes and prediabetes. They had generally low BMIs (eTable 5 in the Supplement), which might be attributed to multiple factors, including economic development and lifestyle factors. According to the survey, Tibetans may have higher levels of physical activity, which is to be expected because of their nomadic way of life. They consume more meat and fewer vegetables or fruits (eTable 5). It has also been reported that Tibetans have a specialized diet, customarily drinking copious tea with salt and butter.²¹ Moreover, persons dwelling at high altitudes have been reported to have lower diabetes prevalence than those living at low altitudes.²² These data are all consistent with findings from this study. It was also previously shown that Tibetan highlanders have accelerated vulnerability to diabetes with lifestyle change or aging.²³ However, reasons for the ethnic variation in diabetes prevalence are difficult to determine from the cross-sectional surveillance data. Further investigation is required to confirm and find the causes.

This study has several strengths. The survey was conducted in a large nationally representative sample of the general population in China, following a strict quality assurance and control program to ensure data validity and reliability. Compared with the 2010 survey, the 2013 survey recruited 70% more individuals and had nearly double the number of study sites, covering more areas within each province and resulting in better representation of the Chinese population. The large sample size of this study provided, to our knowledge, the first direct comparison of diabetes prevalence among major minority groups in China within 1 survey, using the latest diabetes diagnosis criteria.

This study also has several limitations. First, we did not distinguish between type 1 and type 2 diabetes. Limited information is available in regard to the prevalence of type 1 diabetes in China. According to the Chinese clinical guideline for preventing diabetes, less than 5% of patients with diabetes have type 1 diabetes.²⁴ Second, all the glycemic markers in the survey were measured only once for each participant, although the guideline for diagnosing diabetes required confirmation by repeated testing. Third, nonresidents such as college students and migrant workers were not included in the study because of the sampling design. These groups were more likely to be healthy and young, with a lower incidence of diabetes, which could lead to the overestimation of diabetes in the population. Fourth, because of missing data on important information, 9060 participants (5% of the total sample) were excluded from analysis. Although the characteristics of excluded participants were similar (eTable 3), their exclusion could have affected the estimates.

Among adults in China in 2013, the estimated overall prevalence of diabetes was 10.9% and that of prediabetes was 35.7%. Differences from previous estimates for 2010 may be due to an alternate method of measuring HbA_1c.

Corresponding Authors: Linhong Wang, PhD, National Center for Chronic Non-communicable Diseases Control and Prevention, Chinese Center for Disease Control and Prevention, Beijing 100050, China (linhong@chinawch.org.cn); Yonghua Hu, MD, Department of Epidemiology and Biostatistics, School of Public Health, Peking University, Beijing 100191, China (yhhu@bjmu.edu.cn).

Accepted for Publication: May 29, 2017.

Author Contributions: Drs Limin Wang and Linhong Wang had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Limin Wang, Gao, D. Zhang, Deng, Linhong Wang, Hu.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Limin Wang, Gao, D. Zhang, Qin, Jin.

Critical revision of the manuscript for important intellectual content: Limin Wang, Gao, M. Zhang, Huang, Deng, Li, Zhao, Zhou, Tang, Linhong Wang, Hu.

Statistical analysis: Limin Wang, Gao, M. Zhang, Li, Zhao.

Obtained funding: Limin Wang, Gao, Zhou, Linhong Wang, Hu.

Administrative, technical, or material support: Limin Wang, M. Zhang, Huang, D. Zhang, Deng, Li, Zhao, Jin, Zhou, Linhong Wang, Hu.

Supervision: Limin Wang, Gao, Zhou, Linhong Wang, Hu.

Drs Limin Wang and Gao contributed equally as first authors. Drs Linhong Wang and Hu contributed equally as corresponding authors.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Gao reports receiving a grant from the Recruitment Program for Young Professionals in China. No other disclosures were reported.

Funding/Support: This study was supported by the Chinese central government (key project of public health program) and the National Natural Science Foundation of China (grants 81473043, 81230066, and 91546120).

Role of the Funder/Sponsor: None of the funding organizations were involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication.