To investigate the excess mortality and life-years lost associated with diabetes and prediabetes in China.
This national cohort study enrolled 135,405 participants aged 18 years or older from the general population in China. Cox proportional hazards regression models were used to estimate adjusted mortality rate ratio (RR). The life table method was used to estimate life expectancy.
Among the 135,405 participants, 10.5% had diabetes and 36.2% had prediabetes in 2013. During a median follow-up of 6 years, 5517 deaths were recorded, including 1428 and 2300 deaths among people with diabetes and prediabetes, respectively. Diabetes and prediabetes were significantly associated with increased risk of all-cause (diabetes: RR, 1.61 [95% CI 1.49, 1.73]; prediabetes: RR, 1.08 [95% CI 1.01, 1.15]), and cardiovascular disease (diabetes: RR, 1.59 [95% CI 1.41, 1.78]; prediabetes: RR, 1.10 [95% CI 1.00, 1.21]) mortality. Additionally, diabetes was significantly associated with increased risks of death resulting from cancer, respiratory disease, liver disease, and diabetic ketoacidosis or coma. Compared with participants with normoglycemia, life expectancy of those with diabetes and prediabetes was shorter, on average, by 4.2 and 0.7 years at age 40 years, respectively. The magnitude of the associations of diabetes and prediabetes with all-cause and cardiovascular disease mortality varied by age and residence.
In this national study, diabetes and prediabetes were significantly associated with reduced life expectancy and increased all-cause and cause-specific mortality risks. The disparities in excess mortality associated with diabetes and prediabetes between different ages and residences have implications for diabetes and prediabetes prevention and treatment programs.
Introduction
In China, the prevalence of diabetes increased from 0.7% in 1980 to 12.4% in 2018 (1,2). As a result, more than 140 million Chinese adults had diabetes by the end of 2021, and the number of people with diabetes is projected to increase to more than 174 million by 2045 in China (3). In addition to diabetes, the prevalence of prediabetes also has increased rapidly in recent years, reaching 38.1% in 2018 (2).
Diabetes is associated with increased risk of death from cardiovascular diseases (CVD), cancer, and renal diseases and with decreased life expectancy (4). Additionally, prediabetes significantly increased the risk of all-cause death and death from CVD and overall cancer (5–7). In China, CVD and cancer have been the first and second leading causes of excess death, respectively, not only among people with diabetes (8) but also in the general population since the 2000s, and incidence has continued to increase in recent years (9). Due to their high prevalence, diabetes and prediabetes, or hyperglycemia, are considered predominant contributors to the increase in CVD and cancer mortality in China. However, because national data are lacking, the actual contribution of these two diseases to mortality and life-years lost in China remains unknown.
Previous studies have shown that the impacts of diabetes on all-cause and cause-specific mortality could be modified by various sociodemographic factors, such as age, sex, education, income, and residence (4). Identification of these factors has paramount policy implications because policy strategies can be adopted to target high-risk populations.
By using recent, national data, we examined the contribution of diabetes and prediabetes to excess deaths from all causes, and CVD and cancer specifically, in China. We also estimated the effect of diabetes and prediabetes on life expectancy. Moreover, we sought to determine whether the association of diabetes and prediabetes with mortality risk of all-cause deaths, CVD, and cancer varied by sociodemographic factors.
Research Design and Methods
Study Design and Participants
The China Chronic Disease and Risk Factors Surveillance is a nationally representative, cross-sectional, and prospective study organized by the Chinese Center for Disease Control and Prevention. In the 2013 survey, which was conducted from June 2013 to May 2014, 179 347 participants aged 18 years or older were enrolled from 161 sites (n = 97 rural and n = 64 urban locations) across mainland China; the response rate was 93.4%. The multistage sampling and quality control process were reported previously (10). This analysis included 135 405 participants after excluding those with missing information on their identification number, demographic characteristics, major risk factors, or blood measurements (Supplementary Fig. 1). The study protocols were approved by the ethical review committee of the Chinese Center for Disease Control and Prevention and other participating institutions. Written informed consent was obtained from all study participants. Detailed information on data collection and covariates are summarized in the Supplementary Methods.
Assessment of Prediabetes and Diabetes
Diabetes and prediabetes were defined according to American Diabetes Association (ADA) criteria. Diabetes was defined as self-reported prior diagnosis by a health care professional or fasting plasma glucose (FPG) level of 126 mg/dL or higher, 2-h plasma glucose level of 200 mg/dL or higher (after the 75-g glucose tolerance test), or hemoglobin A1c (HbA1c) level of 6.5% or higher. Prediabetes was defined as participants without prior diabetes diagnosis but with a FPG level of 100–125 mg/dL (impaired fasting glucose [IFG]), or a 2-h plasma glucose level of 140–199 mg/dL (impaired glucose tolerance [IGT]), or HbA1c level of 5.7%–6.4%. FPG was measured in all participants after an overnight fast of at least 10 h. FPG level and 2-h plasma glucose level after ingesting 75 g of glucose were measured following the same procedure as reported previously (10,11). HbA1c was measured in a central laboratory with quantitative high-performance liquid chromatography and the boronate affinity method (Bio-Rad D–10 Hemoglobin Analyzer) following a standard protocol (11).
Ascertainment of Death
Death data were ascertained by linking each participant’s name and identification number to the China’s Disease Surveillance Points system through 31 December 2019. All death records were reviewed annually, following quality control procedures (12). Underlying causes of death were classified according to the ICD-10. In addition to all-cause mortality, we classified deaths into 11 cause-specific categories, with each grouping being comprehensive and mutually exclusive. These categories were ischemic heart disease (IHD), stroke, other circulatory disease, respiratory disease, liver disease, renal disease, injuries, infection, diabetes, cancers, and other causes. The Supplementary Table 1 shows specific codes for causes of death.
Statistical Analysis
Sample characteristics were reported as mean ± SD for continuous variables and as numbers with percentages for categorical variables. The proportion of individuals with diabetes and prediabetes among total deaths was determined by calculating the ratio of deaths in these groups to the overall number of deaths. The proportional contribution of leading cause-specific diseases to death was calculated by dividing the number of deaths attributed to each specific disease within each group (normoglycemia, prediabetes, and diabetes) by the total number of deaths within that group. Cox proportional hazards regression models were used to determine the association of baseline prediabetes and diabetes with all-cause and cause-specific mortality. Schoenfeld residuals were used to test the proportional hazards assumption, and no violation was observed. Mortality rate ratios (RRs) and 95% CIs were adjusted for location, age, sex, education level, household income, smoking, alcohol consumption, physical activity, BMI, red meat intake, vegetable and fruit intake, hypertension, dyslipidemia, and self-reported CVD.
Stratified analyses were conducted by age, sex, education, household income, residence, ethnic groups, BMI, smoking status, alcohol consumption, and physical activity. Potential effect modifications were examined by testing the corresponding multiplicative interaction terms.
We quantified the reduction in life expectancy associated with prediabetes and diabetes. Details of the methods used to estimate life expectancy are provided in the Supplementary Methods. By applying Arriaga’s decomposition method, we estimated the cause-specific contributions to the life expectancy difference between people with (pre)diabetes and people with normoglycemia to determine which cause-specific mortality differences were major contributors to the total change in life expectancy (13).
Two-sided P values were used, and P < 0.05 denotes statistical significance. All analyses were performed using SAS, version 9.4 (SAS Institute, Cary, NC). We used Monte Carlo simulation (parametric bootstrapping) with 1,000 runs to calculate the CIs of the life expectancy estimation.
Data Resource and Availability
Because subsequent follow-up investigations are still in progress, data collected for the study, including individual participant data, will not be made available. When all follow-up investigations are finished, data may be made available.
Results
The study enrolled 135,405 participants aged 18 years or older. Of these, 10.5% were adults with diabetes (3.7% previously diagnosed, 6.8% newly diagnosed; 12.1% in urban areas, 9.2% in rural areas; 11.1% men, 9.9% women) and 36.2% were adults with prediabetes (34.5% in urban areas, 37.5% in rural areas; 37.0% men, 35.3% women) (Table 1 and Supplementary Figs. 2 and 3). People with diabetes and prediabetes were older, more likely to be male, and had lower education level and lower household income. They were also more likely to be physically inactive and to have higher levels of alcohol consumption. In addition, people with diabetes and prediabetes were more likely to have higher BMI, higher blood pressure and cholesterol levels, and a history of hypertension and CVD (Table 1). These patterns were similar in urban and rural areas (Supplementary Table 2).
Baseline characteristics of individuals by baseline diabetes status
Characteristic . | People with normoglycemia (n = 63,365) . | People with prediabetes (n = 53,258) . | People with diabetes (n = 18,782) . |
---|---|---|---|
Demographic socioeconomic factors | |||
Age, mean (SD), years | 48.3 (14.0) | 53.9 (13.4) | 57.9 (12.2) |
Female sex | 36,859 (58.2) | 36,261 (56.8) | 10,456 (55.7) |
Urban residence | 28,445 (44.9) | 22,821 (42.9) | 9,697 (51.6) |
Education: junior high school and below | 49,190 (77.6) | 43,875 (82.4) | 15,086 (80.3) |
Household income (<20,000 yuan)* | 12,451 (19.7) | 11,978 (22.5) | 3,924 (20.9) |
Lifestyle factors | |||
Active smoker | 16,196 (25.6) | 13,416 (25.2) | 4,305 (22.9) |
Excessive alcohol drinker | 5,152 (8.1) | 5,294 (9.9) | 1,848 (9.8) |
<150 min/week exercise | 4,999 (7.9) | 4,420 (8.3) | 1,880 (10.0) |
<400 g/day fruits or vegetables | 35,681 (56.3) | 29,814 (56.0) | 10,488 (55.8) |
≥100 g/day red meat | 20,942 (33.1) | 17,432 (32.7) | 5,633 (30.00) |
Anthropometry and blood pressure, mean (SD) | |||
BMI, kg/m2 | 23.6 (3.4) | 24.6 (3.6) | 25.6 (3.8) |
Systolic blood pressure, mmHg | 127.81 (19.5) | 135.18 (20.68) | 142.82 (21.8) |
Diastolic blood pressure, mmHg | 76.24 (11.3) | 78.99 (11.5) | 81.35 (11.7) |
Cholesterol level, mmol/L | 4.7 (1.0) | 5.0 (1.0) | 5.2 (1.1) |
LDL cholesterol, mmol/L | 2.8 (0.8) | 3.1 (0.9) | 3.2 (1.0) |
HDL cholesterol, mmol/L | 1.4 (0.4) | 1.4 (0.4) | 1.3 (0.4) |
Triglycerides, mmol/L | 1.3 (1.0) | 1.6 (1.3) | 2.1 (0.4) |
FPG level, mg/dL | 5.0 (0.4) | 5.8 (0.5) | 8.2 (2.9) |
2-h plasma glucose level, mg/dL | 5.5 (1.1) | 6.8 (1.7) | 11.9 (5.0) |
HbA1c level, % | 5.1 (0.3) | 5.5 (0.4) | 6.8 (1.7) |
Medical history | |||
Hypertension | 18,021 (28.4) | 21,692 (40.7) | 11,085 (59.0) |
CVD | 1,172 (1.9) | 1,518 (2.9) | 1,129 (6.0) |
Cancer | 1,020 (1.6) | 1,049 (2.0) | 505 (2.7) |
Characteristic . | People with normoglycemia (n = 63,365) . | People with prediabetes (n = 53,258) . | People with diabetes (n = 18,782) . |
---|---|---|---|
Demographic socioeconomic factors | |||
Age, mean (SD), years | 48.3 (14.0) | 53.9 (13.4) | 57.9 (12.2) |
Female sex | 36,859 (58.2) | 36,261 (56.8) | 10,456 (55.7) |
Urban residence | 28,445 (44.9) | 22,821 (42.9) | 9,697 (51.6) |
Education: junior high school and below | 49,190 (77.6) | 43,875 (82.4) | 15,086 (80.3) |
Household income (<20,000 yuan)* | 12,451 (19.7) | 11,978 (22.5) | 3,924 (20.9) |
Lifestyle factors | |||
Active smoker | 16,196 (25.6) | 13,416 (25.2) | 4,305 (22.9) |
Excessive alcohol drinker | 5,152 (8.1) | 5,294 (9.9) | 1,848 (9.8) |
<150 min/week exercise | 4,999 (7.9) | 4,420 (8.3) | 1,880 (10.0) |
<400 g/day fruits or vegetables | 35,681 (56.3) | 29,814 (56.0) | 10,488 (55.8) |
≥100 g/day red meat | 20,942 (33.1) | 17,432 (32.7) | 5,633 (30.00) |
Anthropometry and blood pressure, mean (SD) | |||
BMI, kg/m2 | 23.6 (3.4) | 24.6 (3.6) | 25.6 (3.8) |
Systolic blood pressure, mmHg | 127.81 (19.5) | 135.18 (20.68) | 142.82 (21.8) |
Diastolic blood pressure, mmHg | 76.24 (11.3) | 78.99 (11.5) | 81.35 (11.7) |
Cholesterol level, mmol/L | 4.7 (1.0) | 5.0 (1.0) | 5.2 (1.1) |
LDL cholesterol, mmol/L | 2.8 (0.8) | 3.1 (0.9) | 3.2 (1.0) |
HDL cholesterol, mmol/L | 1.4 (0.4) | 1.4 (0.4) | 1.3 (0.4) |
Triglycerides, mmol/L | 1.3 (1.0) | 1.6 (1.3) | 2.1 (0.4) |
FPG level, mg/dL | 5.0 (0.4) | 5.8 (0.5) | 8.2 (2.9) |
2-h plasma glucose level, mg/dL | 5.5 (1.1) | 6.8 (1.7) | 11.9 (5.0) |
HbA1c level, % | 5.1 (0.3) | 5.5 (0.4) | 6.8 (1.7) |
Medical history | |||
Hypertension | 18,021 (28.4) | 21,692 (40.7) | 11,085 (59.0) |
CVD | 1,172 (1.9) | 1,518 (2.9) | 1,129 (6.0) |
Cancer | 1,020 (1.6) | 1,049 (2.0) | 505 (2.7) |
Categorical variables are presented as n (%). Continuous variables are presented as mean ± SD. *There were 31,146 participants who choose not to answer or did not know their household income.
During 801,939 person-years of follow-up (median [interquartile range] follow-up: 6.0 [5.8, 6.3] years), 5,517 participants died (n = 2,300 people with prediabetes and 1,428 people with diabetes). Among total deaths, the proportion of people with diabetes and prediabetes was 25.9% and 41.7%, respectively (Supplementary Fig. 4). CVD and cancer were the first and second cause of death, respectively, among individuals with normal and abnormal blood glucose, accounting for, respectively, 39.4% and 33.9% of the total deaths among people with normoglycemia, 43.3% and 32.3% of the total deaths among people with prediabetes, and 44.7% and 26% of the total deaths among people with diabetes from 2013 to 2019 (Supplementary Fig. 5). Respiratory diseases ranked third among the causes of death among individuals with normoglycemia (9.2%) and prediabetes (9.0%), whereas diabetes ranked third among the causes of death among individuals with diabetes (Supplementary Fig. 5).
The absolute death rate, of people with normoglycemia, prediabetes, and diabetes were 472.94, 731.24, and 1,308.52 per 100,000 person-years, respectively (Table 2). Compared with normoglycemia, diabetes and prediabetes were associated with a 61% (RR, 1.61; 95% CI 1.49, 1.73) and 8% (RR, 1.08; 95% CI 1.01, 1.15) higher risk of all-cause mortality, respectively. The RRs comparing people with diabetes and those with normoglycemia were greater in rural areas than in urban areas (RR: 1.71 [95% CI 1.55, 1.89] vs. 1.50 [95% CI 1.35, 1.68]; P = 0.04 for interaction). When comparing people with prediabetes and those with normoglycemia, the RRs for all-cause mortality were greater at younger ages than older ages (RR: 1.13 [95% CI 1.03, 1.24] vs. 1.03 [95% CI 0.95, 1.13], P = 0.03 for interaction) (Table 3). Prediabetes, defined as IGT and/or elevated HbA1c levels, but not IFG, was associated with an increased risk of all-cause mortality compared with normoglycemia (Supplementary Table 3).
Number of deaths, all-cause and cause-specific mortality rates, and adjusted RRs by diabetes status
. | People with normoglycemia . | People with prediabetes . | People with diabetes . | |||||
---|---|---|---|---|---|---|---|---|
. | No. of deaths . | Deaths per 100,000 person-years . | No. of deaths . | Deaths per 100,000 person-years . | Rate ratio (95% CI) . | No. of deaths . | Deaths per 100,000 person-years . | Rate ratio (95% CI) . |
All-cause | 1,789 | 472.94 | 2,300 | 731.24 | 1.08 (1.01–1.15) | 1,428 | 1,308.52 | 1.61 (1.49–1.73) |
CVD | 704 | 186.11 | 995 | 316.34 | 1.10 (1.00–1.21) | 638 | 584.62 | 1.59 (1.41–1.78) |
IHD | 262 | 69.26 | 375 | 119.22 | 1.13 (0.96–1.33) | 221 | 202.51 | 1.51 (1.25–1.83) |
Stroke | 302 | 79.84 | 405 | 128.76 | 1.05 (0.90–1.22) | 261 | 239.16 | 1.55 (1.30–1.85) |
Other CVD | 140 | 37.01 | 215 | 68.36 | 1.15 (0.93–1.43) | 156 | 142.95 | 1.79 (1.41–2.27) |
Cancers | 607 | 160.47 | 744 | 236.54 | 1.08 (0.97–1.20) | 371 | 339.96 | 1.31 (1.14–1.49) |
Respiratory disease | 165 | 43.62 | 207 | 65.81 | 1.02 (0.82–1.26) | 100 | 91.63 | 1.31 (1.00–1.72) |
Liver disease | 22 | 5.82 | 31 | 9.86 | 1.41 (0.79–2.50) | 23 | 21.08 | 2.93 (1.52–5.63) |
Renal disease | 15 | 3.97 | 29 | 9.22 | 1.68 (0.88–3.21) | 12 | 11 | 1.69 (0.71–4.04) |
Injuries disease | 155 | 40.98 | 142 | 45.15 | 0.93 (0.73–1.17) | 58 | 53.15 | 1.06 (0.77–1.46) |
Infection disease | 36 | 9.52 | 40 | 12.72 | 0.94 (0.59–1.49) | 26 | 23.82 | 1.56 (0.89–2.71) |
Diabetes | 15 | 3.97 | 19 | 6.04 | 1.09 (0.55–2.15) | 132 | 120.96 | 18.95 (10.93–32.84) |
Diabetic ketoacidosis or coma | 4 | 1.06 | 2 | 0.64 | 0.50 (0.09–2.80) | 23 | 21.08 | 17.55 (6.00–51.34) |
Other | 70 | 18.51 | 93 | 29.57 | 1.06 (1.05–1.08) | 68 | 62.31 | 1.09 (1.07–1.11) |
. | People with normoglycemia . | People with prediabetes . | People with diabetes . | |||||
---|---|---|---|---|---|---|---|---|
. | No. of deaths . | Deaths per 100,000 person-years . | No. of deaths . | Deaths per 100,000 person-years . | Rate ratio (95% CI) . | No. of deaths . | Deaths per 100,000 person-years . | Rate ratio (95% CI) . |
All-cause | 1,789 | 472.94 | 2,300 | 731.24 | 1.08 (1.01–1.15) | 1,428 | 1,308.52 | 1.61 (1.49–1.73) |
CVD | 704 | 186.11 | 995 | 316.34 | 1.10 (1.00–1.21) | 638 | 584.62 | 1.59 (1.41–1.78) |
IHD | 262 | 69.26 | 375 | 119.22 | 1.13 (0.96–1.33) | 221 | 202.51 | 1.51 (1.25–1.83) |
Stroke | 302 | 79.84 | 405 | 128.76 | 1.05 (0.90–1.22) | 261 | 239.16 | 1.55 (1.30–1.85) |
Other CVD | 140 | 37.01 | 215 | 68.36 | 1.15 (0.93–1.43) | 156 | 142.95 | 1.79 (1.41–2.27) |
Cancers | 607 | 160.47 | 744 | 236.54 | 1.08 (0.97–1.20) | 371 | 339.96 | 1.31 (1.14–1.49) |
Respiratory disease | 165 | 43.62 | 207 | 65.81 | 1.02 (0.82–1.26) | 100 | 91.63 | 1.31 (1.00–1.72) |
Liver disease | 22 | 5.82 | 31 | 9.86 | 1.41 (0.79–2.50) | 23 | 21.08 | 2.93 (1.52–5.63) |
Renal disease | 15 | 3.97 | 29 | 9.22 | 1.68 (0.88–3.21) | 12 | 11 | 1.69 (0.71–4.04) |
Injuries disease | 155 | 40.98 | 142 | 45.15 | 0.93 (0.73–1.17) | 58 | 53.15 | 1.06 (0.77–1.46) |
Infection disease | 36 | 9.52 | 40 | 12.72 | 0.94 (0.59–1.49) | 26 | 23.82 | 1.56 (0.89–2.71) |
Diabetes | 15 | 3.97 | 19 | 6.04 | 1.09 (0.55–2.15) | 132 | 120.96 | 18.95 (10.93–32.84) |
Diabetic ketoacidosis or coma | 4 | 1.06 | 2 | 0.64 | 0.50 (0.09–2.80) | 23 | 21.08 | 17.55 (6.00–51.34) |
Other | 70 | 18.51 | 93 | 29.57 | 1.06 (1.05–1.08) | 68 | 62.31 | 1.09 (1.07–1.11) |
The RRs were adjusted for age (5-year group), sex (male, or female), residence (urban, or rural), education (junior high school and below, high school, college and above), household income (<¥20,000/year, ≥¥20,000/year, or no answer/did not know), smoking (never smoked, past smoker, or active smoker), alcohol consumption (excessive, rare/occasional, or nondrinker), physical activity (<150 min/week or ≥150 min/week), red meat intake (<100 g/day or ≥100 g/day), vegetable and fruit intake (<400 g/day or ≥400 g/day), BMI (<18.5, 18.5–23.9, 24–27.9, ≥28 kg/m2), hypertension (yes or no), dyslipidemia (yes or no) and self-reported CVD (yes or no).
Number of deaths, mortality rates, and adjusted RRs for all-cause and cause-specific mortality by diabetes status and age, sex, or residence area
. | People with normoglycemia . | People with prediabetes . | People with diabetes . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
. | No. of deaths . | Deaths per 100,000 person-years . | No. of deaths . | Deaths per 100,000 person-years . | RR (95% CI) . | P for interaction . | No. of deaths . | Deaths per 100,000 person-years . | RR (95% CI) . | P for interaction . | |
All-cause | 0.03 | 0.13 | |||||||||
Age, years | |||||||||||
<65 | 862 | 257.31 | 925 | 366.69 | 1.13 (1.03–1.24) | 534 | 673.71 | 1.75 (1.56–1.96) | |||
≥65 | 927 | 2,142.52 | 1,375 | 2,207.95 | 1.03 (0.95–1.13) | 894 | 2,993.11 | 1.50 (1.36–1.65) | |||
0.45 | 0.77 | ||||||||||
Sex | |||||||||||
Male | 1,038 | 658.9 | 1,329 | 984.9 | 1.12 (1.03–1.21) | 771 | 1,605.9 | 1.61 (1.46–1.78) | |||
Female | 751 | 340.22 | 971 | 540.66 | 1.02 (0.93–1.12) | 657 | 1,074.93 | 1.59 (1.42–1.78) | |||
0.45 | 0.04 | ||||||||||
Residence | |||||||||||
Rural | 1,074 | 503.53 | 1,366 | 775.35 | 1.10 (1.02–1.20) | 697 | 1,411.37 | 1.71 (1.55–1.89) | |||
Urban | 715 | 433.39 | 934 | 675.07 | 1.05 (0.95–1.16) | 731 | 1,223.51 | 1.50 (1.35–1.68) | |||
CVD | 0.01 | 0.002 | |||||||||
Age, years | |||||||||||
<65 | 260 | 77.61 | 322 | 127.65 | 1.20 (1.02–1.42) | 215 | 271.25 | 1.89 (1.56–2.30) | |||
≥65 | 444 | 1,026.19 | 673 | 1,080.69 | 1.03 (0.91–1.16) | 423 | 1,416.2 | 1.42 (1.23–1.63) | |||
0.43 | 0.47 | ||||||||||
Sex | |||||||||||
Male | 392 | 248.83 | 552 | 409.08 | 1.15 (1.01–1.31) | 311 | 647.78 | 1.49 (1.27–1.74) | |||
Female | 312 | 141.34 | 443 | 246.66 | 1.03 (0.89–1.19) | 327 | 535.01 | 1.66 (1.41–1.96) | |||
0.05 | 0.008 | ||||||||||
Residence | |||||||||||
Rural | 412 | 193.16 | 608 | 345.11 | 1.18 (1.04–1.34) | 319 | 645.95 | 1.75 (1.50–2.04) | |||
Urban | 292 | 176.99 | 387 | 279.71 | 0.98 (0.84–1.15) | 319 | 533.93 | 1.41 (1.19–1.67) | |||
Cancer | 0.48 | 0.50 | |||||||||
Age, years | |||||||||||
<65 | 364 | 108.65 | 380 | 150.64 | 1.10 (0.95–1.28) | 164 | 206.91 | 1.33 (1.09–1.61) | |||
≥65 | 243 | 561.63 | 364 | 584.5 | 1.05 (0.89–1.24) | 207 | 693.04 | 1.26 (1.04–1.53) | |||
0.87 | 0.61 | ||||||||||
Sex | |||||||||||
Male | 367 | 232.96 | 445 | 329.78 | 1.09 (0.95–1.25) | 220 | 458.23 | 1.35 (1.14–1.62) | |||
Female | 240 | 108.73 | 299 | 166.48 | 1.06 (0.89–1.26) | 151 | 247.05 | 1.25 (0.99–1.54) | |||
0.15 | 0.36 | ||||||||||
Residence | |||||||||||
Rural | 368 | 172.53 | 409 | 232.15 | 1.03 (0.89–1.18) | 160 | 323.99 | 1.27 (1.04–1.55) | |||
Urban | 239 | 144.87 | 335 | 242.13 | 1.16 (0.98–1.38) | 211 | 353.16 | 1.35 (1.11–1.65) |
. | People with normoglycemia . | People with prediabetes . | People with diabetes . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
. | No. of deaths . | Deaths per 100,000 person-years . | No. of deaths . | Deaths per 100,000 person-years . | RR (95% CI) . | P for interaction . | No. of deaths . | Deaths per 100,000 person-years . | RR (95% CI) . | P for interaction . | |
All-cause | 0.03 | 0.13 | |||||||||
Age, years | |||||||||||
<65 | 862 | 257.31 | 925 | 366.69 | 1.13 (1.03–1.24) | 534 | 673.71 | 1.75 (1.56–1.96) | |||
≥65 | 927 | 2,142.52 | 1,375 | 2,207.95 | 1.03 (0.95–1.13) | 894 | 2,993.11 | 1.50 (1.36–1.65) | |||
0.45 | 0.77 | ||||||||||
Sex | |||||||||||
Male | 1,038 | 658.9 | 1,329 | 984.9 | 1.12 (1.03–1.21) | 771 | 1,605.9 | 1.61 (1.46–1.78) | |||
Female | 751 | 340.22 | 971 | 540.66 | 1.02 (0.93–1.12) | 657 | 1,074.93 | 1.59 (1.42–1.78) | |||
0.45 | 0.04 | ||||||||||
Residence | |||||||||||
Rural | 1,074 | 503.53 | 1,366 | 775.35 | 1.10 (1.02–1.20) | 697 | 1,411.37 | 1.71 (1.55–1.89) | |||
Urban | 715 | 433.39 | 934 | 675.07 | 1.05 (0.95–1.16) | 731 | 1,223.51 | 1.50 (1.35–1.68) | |||
CVD | 0.01 | 0.002 | |||||||||
Age, years | |||||||||||
<65 | 260 | 77.61 | 322 | 127.65 | 1.20 (1.02–1.42) | 215 | 271.25 | 1.89 (1.56–2.30) | |||
≥65 | 444 | 1,026.19 | 673 | 1,080.69 | 1.03 (0.91–1.16) | 423 | 1,416.2 | 1.42 (1.23–1.63) | |||
0.43 | 0.47 | ||||||||||
Sex | |||||||||||
Male | 392 | 248.83 | 552 | 409.08 | 1.15 (1.01–1.31) | 311 | 647.78 | 1.49 (1.27–1.74) | |||
Female | 312 | 141.34 | 443 | 246.66 | 1.03 (0.89–1.19) | 327 | 535.01 | 1.66 (1.41–1.96) | |||
0.05 | 0.008 | ||||||||||
Residence | |||||||||||
Rural | 412 | 193.16 | 608 | 345.11 | 1.18 (1.04–1.34) | 319 | 645.95 | 1.75 (1.50–2.04) | |||
Urban | 292 | 176.99 | 387 | 279.71 | 0.98 (0.84–1.15) | 319 | 533.93 | 1.41 (1.19–1.67) | |||
Cancer | 0.48 | 0.50 | |||||||||
Age, years | |||||||||||
<65 | 364 | 108.65 | 380 | 150.64 | 1.10 (0.95–1.28) | 164 | 206.91 | 1.33 (1.09–1.61) | |||
≥65 | 243 | 561.63 | 364 | 584.5 | 1.05 (0.89–1.24) | 207 | 693.04 | 1.26 (1.04–1.53) | |||
0.87 | 0.61 | ||||||||||
Sex | |||||||||||
Male | 367 | 232.96 | 445 | 329.78 | 1.09 (0.95–1.25) | 220 | 458.23 | 1.35 (1.14–1.62) | |||
Female | 240 | 108.73 | 299 | 166.48 | 1.06 (0.89–1.26) | 151 | 247.05 | 1.25 (0.99–1.54) | |||
0.15 | 0.36 | ||||||||||
Residence | |||||||||||
Rural | 368 | 172.53 | 409 | 232.15 | 1.03 (0.89–1.18) | 160 | 323.99 | 1.27 (1.04–1.55) | |||
Urban | 239 | 144.87 | 335 | 242.13 | 1.16 (0.98–1.38) | 211 | 353.16 | 1.35 (1.11–1.65) |
The RRs were adjusted for age (5-year group), sex (male or female), residence (urban or rural), education (junior high school and below, high school, college and above), household income (<¥20,000/year, ≥¥20,000/year, or not answer/did not know), smoking (never smoked, past smoker, or active smoker), alcohol consumption (excessive, rare/occasional, or nondrinker), physical activity (<150 min/week or ≥150 min/week), red meat intake (<100 g/day or ≥100 g/day), vegetable and fruit intake (<400 g/day or ≥400 g/day), BMI (<18.5, 18.5–23.9, 24–27.9, ≥28 kg/m2), hypertension (yes or no), dyslipidemia (yes or no), and self-reported CVD (yes or no). The strata variable was not included in the model when stratifying by itself, except for age.
Life expectancy for people aged 40 years with normoglycemia, prediabetes, and diabetes was 38.2 years (95% CI 38.0, 38.4), 37.5 years (95% CI 37.2, 37.8), and 33.9 years (95% CI 33.6, 34.2) for men, and 43.3 years (95% CI 43.0, 43.5), 42.6 years (95% CI 42.3, 42.8), and 39.1 years (95% CI 38.7, 39.4) for women (Fig. 1A and B), respectively. Life expectancy estimates according to diabetes status at ages 60 and 80 years are also presented in Fig. 1A and B. Compared with those with normoglycemia, men and women with diabetes at age 40 years died 4.3 years (95% CI 4.2, 4.5) and 4.2 years (95% CI 4.1, 4.4) earlier, respectively. At age 40 years, both men and women with prediabetes died 0.7 years (95% CI 0.6, 0.8) earlier. (Fig. 1C and D). Moreover, CVD deaths were the leading cause (accounting for about 40% or more) of the reduction in life expectancy associated with diabetes or prediabetes, with the remaining percentage being largely due to non-CVD, and non-cancer deaths in diabetes, or overall cancer in prediabetes (Fig. 1E–H).
Life expectancy and years of life lost by diabetes status. A and B: Estimated life expectancy at age 40, 60, and 80 years by diabetes status. C and D: Loss of life expectancy of people with prediabetes or diabetes. E–H: Estimated years of life lost attributable to increased deaths from cardiovascular disease, cancer, and other causes in people with prediabetes or diabetes.
Life expectancy and years of life lost by diabetes status. A and B: Estimated life expectancy at age 40, 60, and 80 years by diabetes status. C and D: Loss of life expectancy of people with prediabetes or diabetes. E–H: Estimated years of life lost attributable to increased deaths from cardiovascular disease, cancer, and other causes in people with prediabetes or diabetes.
The absolute CVD death rates of people with normoglycemia, prediabetes, and diabetes were 186.11, 316.34, and 584.62 per 100,000 person-years, respectively. Compared with normoglycemia, diabetes and prediabetes were associated with a 59% (RR, 1.59; 95% CI 1.41, 1.78) and 10% (RR, 1.10; 95% CI 1.00, 1.21) higher risk of CVD mortality, respectively (Table 2). In people with diabetes, the RRs for CVD mortality were greater in rural areas than in urban areas (RR: 1.75 [95% CI 1.50, 2.04] vs. 1.41 [95% CI 1.19, 1.67]; P = 0.008 for interaction), and greater at younger ages than at older ages (RR: 1.89 [95% CI 1.56, 2.30] vs. 1.42 [95% CI 1.23, 1.63]; P = 0.002 for interaction) (Table 3). In people with prediabetes, similar differences in CVD mortality risk were also observed in resident areas (RR: 1.18 [95% CI 1.04, 1.34] for rural areas vs. 0.98 [95% CI 0.84, 1.15] for urban areas, P = 0.05 for interaction) and age (RR: 1.20 [95% CI 1.02–1.42] for younger ages vs. 1.03 [95% CI 0.91, 1.16] for older ages; P = 0.01 for interaction) (Table 3). For the subtypes of CVD mortality, the adjusted RRs comparing diabetes and prediabetes with normoglycemia were 1.51 (95% CI 1.25, 1.83) and 1.13 (95% CI 0.96, 1.33) for IHD; 1.55 (95% CI 1.30, 1.85) and 1.05 (95% CI 0.90, 1.22) for stroke; 1.79 (95% CI 1.41, 2.27); and 1.15 (95% CI 0.93, 1.43) for other circulatory diseases, respectively (Table 2). In both men and women, we found higher risk of IHD mortality associated with diabetes and prediabetes in people aged <65 years than those aged ≥65 years, with a significant interaction (P ≤ 0.001 for diabetes and P = 0.04 for prediabetes, for interaction) (Supplementary Table 4). Women with diabetes who were younger than 65 years had a 191% increased risk of IHD mortality compared with normoglycemic women (2.91; 95% CI 1.80, 4.71), and men in the same age-group with diabetes had a 77% increased risk (1.77; 95% CI 1.14, 2.74) (Supplementary Table 4). The results of stratified analyses by age, sex, and residence area are listed in Supplementary Table 5.
The absolute overall cancer death rates of people with normoglycemia, prediabetes, and diabetes were 160.47, 236.54, and 339.96 per 100,000 person-years, respectively. Compared with normoglycemia, diabetes was associated with a 31% (RR, 1.31; [95% CI 1.14, 1.49]) higher risk of overall cancer mortality (Table 2). Specifically, compared with normoglycemia, diabetes was associated with increased RRs for death from lung cancer (RR, 1.40; 95% CI 1.09, 1.81), and liver cancer (RR, 2.34; 95% CI 1.68, 3.26) (Supplementary Table 6). Prediabetes was not significantly associated with higher risk of cancer mortality. The RRs for overall cancer mortality were not significantly modified by age, sex, and resident areas (Table 3).
As for the other diseases, diabetes was associated with a significantly elevated RR of 17.55 (95% CI 6.00, 51.34) for death due to diabetic ketoacidosis or coma (Table 2). In addition, diabetes was associated with elevated RRs of 2.93 (95% CI 1.52, 5.63) for death resulting from liver disease and of 1.31 (95% CI 1.00, 1.72) for death resulting from respiratory disease (Table 2). The associations of diabetes and prediabetes with the risk of all-cause, CVD, and cancer deaths were similar across categories of several other baseline characteristics, including household income, education level, smoking status, and alcohol consumption (Supplementary Tables 7–9). A significant interaction between BMI and diabetes status relative to all-cause mortality was observed, with higher mortality risk associated with diabetes observed in those with BMI<18.5 kg/m2 (Supplementary Tables 7–9). We conducted a comparison of baseline characteristics between participants with and without identification numbers and did not detect any significant differences between two groups (Supplementary Table 10).
Conclusions
In this national longitudinal study, both diabetes and prediabetes were associated with an increased risk of all-cause mortality and CVD mortality, and a reduction in life expectancy in China. Additionally, diabetes was associated with increased risks of death from cancer, respiratory disease, liver disease, and diabetic ketoacidosis or coma. Notably, the associations of diabetes and prediabetes with all-cause and CVD deaths were modified by age and residence.
To our knowledge, no previous studies have provided estimates of reduction in life expectancy associated with diabetes and prediabetes in China. By using these contemporary mortality data and a national cohort, we found that the life expectancy of people with diabetes and prediabetes at age 40 years was, on average, 4.2 years less than that of people with normoglycemia. This diabetes-related life expectancy loss was similar to or slightly higher than that in many high-income countries, such as Finland (4.0 years for men and 3.2 years for women in 2016–2017) and Australia (4.1 years for men and 3.9 years for women in 2018–2019) (14). In addition, the life expectancy of people with prediabetes at age 40 years was, on average, 0.7 years less than that of people with normoglycemia. Although this impact was weaker than that of diabetes, the high prevalence of prediabetes (approximately 36% in 2013) (1) underscores its significant contribution to the reduced life expectancy of the overall population in China. Notably, because information about glycemic status was obtained at baseline, it remains uncertain how many people with normoglycemia or prediabetes at baseline progressed to (pre)diabetes during follow-up. Therefore, we calculated the transition probability (14) and found a slight difference in estimate of years of life lost (approximately 1% lower for men and 2% lower for women). CVD and cancer were two major causes of death among people with diabetes and prediabetes (15). In this study, diabetes conferred a 59% increase in CVD mortality risk and a 31% increase in cancer mortality risk. Prediabetes was associated a 10% increase in CVD mortality risk but showed no association with cancer mortality risk. These results are generally consistent with the data reported in some Western countries (4,16,17). For example, a study conducted in the U.S. showed that the RRs for CVD and cancer mortality were 1.79 (95% CI 1.42, 2.27) and 1.11 (95% CI 0.88, 1.41), respectively, associated with diabetes; and 1.18 (95% CI 0.93, 1.51) and 1.09 (95% CI 0.88, 1.35), respectively, associated with prediabetes (18). In England, it was reported that the RRs for CVD and cancer associated with diabetes were 1.99 (95% CI 1.79, 2.21) and 1.30 (95% CI 1.18, 1.43), respectively (17). Notably, the CVD and cancer mortality rates in people with diabetes in the present study are almost identical to those reported from another Chinese cohort followed from 2004 to 2013 (CVD, 584 vs. 538 per 100,000 person-years; cancer, 339 vs. 300 per 100,000 person-years) (8), despite population characteristics (e.g., age, CVD history, blood pressure) being different between the two studies. This is in contrast to many high-income countries, which have had a decline in the diabetes-related mortality rates of CVD and cancer since the 1990s (4,19). Often, the reduction in high-income countries is attributed to the nationwide control of risk factors, such as blood glucose, LDL cholesterol, blood pressure, and smoking (16). On the other hand, by ensuring access to diabetes care and patient education, these countries effectively improved the treatment and control of diabetes in the population (e.g., >80% of the population with diabetes in the U.S. used glucose-lowering medication, and >70% of those had an HbA1c level lower than 7% (20). However, in China, a huge number of patients have undiagnosed, untreated, or uncontrolled diabetes: the incidence of awareness and treatment of diabetes among the population were 36.7% and 32.9%, respectively, in 2018 (1,21).
In this study, we found that IGT and elevated HbA1c levels were associated with increased all-cause mortality, whereas IFG was not. To date, the associations between different prediabetes definitions based on the ADA criteria with all-cause mortality are controversial. One meta-analysis also showed that both IFG and IGT correlated well with increased all-cause mortality, whereas elevated HbA1c levels did not; moreover, the IFG-associated increase in all-cause mortality risk was mainly driven by FPG in the range of 110–125 mg/dL (5). However, in this study, 110–125 mg/dL FPG was not associated with increased all-cause mortality. In an umbrella analysis, IFG and IGT , but not elevated HbA1c levels, were associated with the increased RR of all-cause mortality (22). However, in a Japanese study, both IFG and elevated HbA1c levels were associated with all-cause mortality (23). Therefore, more high-quality, large-scale studies are required to clarify the association between different prediabetes criteria and mortality risk in the Chinese population.
In this study, 44.7% of people with diabetes died of CVD, which is higher than average proportion (30%) around the world in 2019 (4). However, it is possible that the lower percentage of deaths due to CVD in other countries could be attributed to the reduced CVD mortality among people with diabetes who are older than 80 years (24), which is typical in high-income nations with older populations. In contrast, the relatively young participants (average age, 57.8 years) in our study might contribute to the higher percentage of deaths due to CVD. However, only 6% of people with diabetes had self-reported CVD history at baseline in this study. This result was in line with the report of China National HbA1c Surveillance System, which showed that the combined prevalence of diagnosed coronary heart disease and stroke was 5.85% among urban residents with diabetes in 2012 (25), indicating a low awareness of diagnosed CVD among individuals with diabetes. On the other hand, although it has been reported that the prevalence of diabetes is about 35–50% among patients with CVD in China (26,27), the awareness of diabetes among patients with CVD overall was still very low (27) because oral glucose tolerance tests and HbA1c have not been routinely performed in the clinical workup to assess the diabetes status of patients with CVD in most Chinese hospitals (26). Given that individuals with both diabetes and CVD are at extremely high risk of developing fatal CVD events, screening for CVD and related risk factors among people with diabetes and screening for diabetes among patients with CVD should be routinely performed in primary health care.
In China, the CVD mortality rate in rural areas has been higher than in urban areas since 2013 (323 vs. 278 per 100,000 person-years in 2019) (9). In line with this, diabetes and prediabetes conferred higher CVD mortality risks in rural areas than in urban areas (645.95 vs. 533.93 per 100,000 person-years and 345.11 vs. 279.71 per 100,000 person-years, respectively). Because people with diabetes and prediabetes account for the largest proportion of population who die of CVD in China, these data indicate that hyperglycemia is a major cause of rural-urban disparity in CVD mortality rate. Therefore, efforts to enhance prevention and treatment of diabetes, prediabetes, and related CVD complications in rural areas should be prioritized to reduce the rural-urban disparity in CVD mortality rate.
In this study, we found that men and women with diabetes who were younger than 65 years had, respectively, a 77% and 191% increased risk of IHD mortality than normoglycemic people (28–31). The higher risk among women is often attributed to the attenuation of cardiovascular protection associated with female sex, although the underlying mechanisms still need to be explored (32). On the other hand, younger people with diabetes also face a higher relative risk of all-cause and CVD mortality compared with older people (7,33–35). This is because younger people with diabetes have longer disease exposure, which leads to increased risk for chronic complications and adverse societal impact of the disease (36). Despite relative risks being higher in younger people with diabetes, older people with diabetes face a significantly higher absolute risk (37).
Among individuals with diabetes in our study, the mean systolic blood pressure was 143 mmHg, which is consistent with reports from the U.S. and the U.K. (mean, 140 mmHg) (38,39). Furthermore, the HbA1c levels among individuals with normoglycemia, prediabetes, and diabetes in our study were 5.1%, 5.5%, and 6.8%, respectively, which are slightly lower than those reported in the Atherosclerosis Risk in Communities study (5.3%, 6.0%, and 7.4%, respectively) (40). Additionally, despite the slightly lower HbA1c levels and comparable mean systolic blood pressure observed in people with diabetes in the present study, the awareness and treatment of diabetes and hypertension remained low in China, highlighting an urgent need for enhanced glycemic and blood pressure management strategies among people with prediabetes and diabetes in China.
A major strength of this study is that we used a large national sample of the general population in China, following a strict quality assurance and control protocol to ensure data validity and reliability. Other strengths include a long-term follow-up, a near-complete ascertainment of overall and cause-specific mortality, and detailed assessment of covariates. Several limitations of this study need to be acknowledged. First, the relatively few deaths of patients with some types of cancer limited the power to provide precise estimates. Second, no detailed information was available about the medications used, diabetes duration, and complications among people with diabetes and prediabetes. Third, we did not collect adequate information on diabetes type, although more than 90–95% of cases are likely to be type 2 diabetes. Fourth, due to the unavailability of detailed death data, there may be an underestimation of CVD deaths among individuals with diabetes. Fifth, information on glycemic status was only obtained at one time point, and changes in glycemic status over time were not captured in the present analysis. Consequently, our estimates of the years of life lost might not accurately reflect the true disease impact for the whole population over time. Future studies with serial measurements of glycemic status, including trajectory analyses, are needed. Sixth, we excluded participants without identification numbers, due to the inability to link them with the mortality database. This may introduce selection bias, although baseline characteristics showed no significant differences between participants with and without identification numbers. Finally, due to the observational nature of the study, the causality cannot be determined.
Conclusion
Diabetes and prediabetes were associated with increased risk of all-cause and CVD mortality, and a reduction in life expectancy in China. Additionally, diabetes was associated with increased risk of death from cancer, liver disease, respiratory disease, and diabetic ketoacidosis or coma. Moreover, there were age and residence differences in all-cause and CVD mortality associated with diabetes and prediabetes. More research is needed to better understand these disparities and to develop targeted interventions to reduce the burden of diabetes and its complications in China.
This article contains supplementary material online at https://doi.org/10.2337/figshare.26814685.
This article is featured in a podcast available at diabetesjournals.org/journals/pages/diabetes-core-update-podcasts.
Article Information
Acknowledgments. We thank Dr. Frank Hu, of Harvard University, for his constructive comments on the manuscript.
Funding. This study was supported by the National Key R&D Program of China (research grant 2023YFC2506504).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. K.H., G.L., and M.Z. jointly directed this work, conceived and designed the study, and take responsibility for the integrity of the data and the accuracy of the data analysis. K.H., G.L., Y.T., and Z.Q. drafted the manuscript. Z.Q., Y.T., F.W., S.D., and G.L. conducted the analysis. Y.T., Z.Q., F.W., S.D., and Y.W. completed the follow-up work. Y.T., Z.Q., F.W., S.D., G.L., Y.W., and Z.W. accessed and verified the underlying data. P.Y. and Y.H. provided administrative, technical, or material support. All authors critically revised the manuscript for important intellectual content and gave final approval for the version to be published; agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved; and had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Handling Editors. The journal editors responsible for overseeing the review of the manuscript were Elizabeth Selvin and Jonathan E. Shaw.