East Asia and high-income Asia Pacific burden of diabetes mellitus from 1990 to 2023

Summary

Diabetes mellitus (DM) remains a major health challenge in East Asia and the high-income Asia Pacific, yet systematic subtype-specific comparisons are lacking. Using Global Burden of Disease Study 2023 data, we analyzed age-standardized incidence, prevalence, mortality, and disability-adjusted life year (DALY) rates for DM subtypes from 1990 to 2023, calculated mortality-to-incidence ratios (MIRs) and population-attributable fractions, and projected outcomes to 2050. In 2023, East Asia had higher MIR (0.04) and mortality (8.26/100,000) than the high-income Asia Pacific (0.02 and 4.50/100,000), with type 2 diabetes (T2DM) comprising >94% of cases. High body-mass index was the primary T2DM risk factor (49.3%); T1DM burden was linked to temperature exposure. Under the reference scenario, T2DM mortality and DALY rates will rise by 2050, but optimized behavioral and metabolic interventions could reduce DM DALY rates by >84% in East Asia and >70% in the high-income Asia Pacific, highlighting the urgent need for targeted, risk-factor-specific diabetes control strategies.

Graphical abstract

Highlights

• •East Asia has a higher DM disability burden than the High-income Asia Pacific
• •MIR disparities reveal regional gaps in diabetes care
• •High BMI dominated T2DM risk; extreme temperatures affected T1DM
• •Targeted interventions could cut DM burden sharply by 2050

Health sciences; Medicine; Medical specialty; Internal medicine; Endocrinology; Public health

Introduction

Diabetes mellitus (DM) is an endocrine and metabolic disorder characterized by chronic hyperglycemia caused by defective insulin secretion or impaired biological action of insulin.^1,2,3 As a complex lifelong clinical condition, the adverse impacts of DM extend far beyond glycemic abnormalities. Its multiple systemic complications, including cardiovascular and cerebrovascular diseases, renal failure, retinopathy, and peripheral nerve damage, substantially drive high disability and mortality rates, rendering DM one of the most formidable public health challenges worldwide.^4,5 The global prevalence of DM continues to expand at an alarming pace. According to the latest data released by the International Diabetes Federation (IDF), 589 million adults aged 20–79 years were living with DM globally in 2024, equivalent to approximately one in every 11 adults. In the same year, DM and its related complications contributed to 6.7 million deaths worldwide, underscoring its substantial lethal burden.⁶ Alarmingly, the IDF projects that the global number of patients with DM will rise to 853 million by 2050. Such a sustained rapid growth trend is poised to impose unprecedented pressure on global healthcare systems in the coming decades.⁶

Against this severe public health backdrop, a comprehensive understanding of DM disease burden requires etiological differentiation between its two major subtypes: type 1 diabetes mellitus (T1DM) and type 2 diabetes mellitus (T2DM). T1DM is predominantly driven by autoimmune-mediated destruction of pancreatic β-cells, which leads to absolute insulin deficiency. Accounting for approximately 5%–10% of all DM cases, T1DM primarily affects children and adolescents.⁷ By contrast, T2DM arises mainly from insulin resistance combined with progressive impairment of β-cell function, resulting in relative insulin insufficiency. Closely linked to obesity, genetic predisposition, and population aging, T2DM accounts for over 90% of the global DM burden and predominates in middle-aged and elderly populations.^8,9 These two subtypes differ fundamentally in etiology, age of onset, clinical management, and long-term prognosis, thereby presenting distinct disease burden patterns and evolutionary trajectories. Therefore, precise estimation of the independent disease burden attributed to each subtype is essential for formulating targeted prevention and control strategies.

As one of the most densely populated regions worldwide, East Asia bears an extraordinarily heavy DM burden. China alone accounts for roughly one-quarter of the global DM patient population.¹⁰ Its prevalence has surged sharply from 2.5% in 1994 to 13.0% in 2021, with a continued increasing trend projected in the future.¹¹ Meanwhile, despite a higher level of economic development, the High-income Asia Pacific also faces an accelerating DM epidemic driven by rapid population aging. For instance, DM has become a leading cause of non-communicable disease burden in the Republic of Korea, whereas existing public health interventions remain inadequate relative to the severity of the epidemic¹² Against this backdrop, systematic exploration of DM epidemic trends, population distribution features, risk factor attribution, and inter-regional disparities between East Asia and High-income Asia Pacific is of urgent practical significance. The findings can support the optimization of prevention and control strategies, rational allocation of health resources, and advancement of integrated health-centered service systems.¹³

Although numerous studies have explored DM disease burden based on the Global Burden of Disease (GBD) framework, most existing research focuses only on global overall trends or a single DM subtype. Few studies have systematically compared East Asia and High-income Asia Pacific, two regions with distinct economic development levels, population structures, and diabetes epidemic profiles. Furthermore, most previous studies fail to independently delineate the temporal burden trajectories of T1DM and T2DM separately, and lack cross-regional comparisons of the mortality-to-incidence ratio (MIR), a key indicator reflecting the disease burden paradox. More importantly, GBD 2023 has incorporated updated data sources, refined analytical models such as DisMod-AT, and newly established risk-outcome pairs. These updates render previous conclusions derived from older GBD versions, necessitating systematic reassessment. Accordingly, based on the GBD 2023, this study performed a systematic evaluation of DM and its subtype burden at both regional and national levels across East Asia and High-income Asia Pacific. We quantified temporal trends in age-standardized incidence, prevalence, mortality, and disability-adjusted life years (DALYs) for T1DM and T2DM from 1990 to 2023. The MIR was calculated to characterize the disease burden paradox, and population-attributable burdens of major risk factors were evaluated. We further projected future epidemic trends up to 2050. Ultimately, this study aims to provide robust evidence-based references for the formulation of precise, subtype-specific prevention and control measures, as well as long-term healthcare resource planning across East Asia, High-income Asia Pacific, and individual constituent countries.

Results

East Asia bears a higher diabetes mortality burden than the high-income Asia Pacific in 2023, driven by T2DM and marked by wide international disparities

In 2023, DM and its major subtypes posed significant public health challenges in East Asia and High-income Asia Pacific, with marked geographical inequities in disease burden.

Overall DM: East Asia has a lower incidence but substantially higher ASMR (8.26 vs. 4.50 per 100,000) and ASDR

In East Asia, the ASIR of DM was 217.71 per 100,000 (95% UI: 199.53–235.95), and the ASPR was 5477.91 per 100,000 (95% UI: 5034.65–6003.23). The mortality and DALYs burden was relatively heavy, with ASMR of 8.26 per 100,000 (95% UI: 6.58–10.02) and ASDR of 535.40 per 100,000 (95% UI: 411.42–679.41). In High-income Asia Pacific, the ASIR was 266.50 per 100,000 (95% UI: 247.61–287.32), and the ASPR was 5341.92 per 100,000 (95% UI: 4920.36–5766.25). The mortality and DALYs burden was also severe, with ASMR of 4.50 per 100,000 (95% UI: 3.81–5.24) and ASDR of 519.92 per 100,000 (95% UI: 382.34–668.19) (Table 1).

Table 1.

Age-standardized rates of diabetes incidence, prevalence, mortality, and disability-adjusted life years (DALYs) in 1990 and 2023, with estimated total percentage change (TPC) from 1990 to 2023, stratified by diabetes type and geographic region

Cause	location	Incidence			Prevalence			Mortality			DALYs
		ASIR (1/100000) (95% UI)	ASIR (1/100000) (95% UI)	TPC (95% UI)	ASPR (1/100000) (95% UI)	ASPR (1/100000) (95% UI)	TPC (95% UI)	ASMR (1/100000) (95% UI)	ASMR (1/100000) (95% UI)	TPC (95% UI)	ASDR (1/100000) (95% UI)	ASDR (1/100000) (95% UI)	TPC (95% UI)
		1990	2023	1990–2023	1990	2023	1990–2023	1990	2023	1990–2023	1990	2023	1990–2023
DM	East Asia	179.55 (161.44–197.52)	217.71 (199.53–235.95)	0.21 (0.18–0.25)	4405.60 (3950.60–4923.79)	5477.91 (5034.65–6003.23)	0.24 (0.21–0.29)	9.97 (7.69–12.70)	8.26 (6.58–10.02)	−0.17 (−0.36− 0.14)	526.71 (411.48–644.55)	535.40 (411.42–679.41)	0.02 (−0.10− 0.14)
	China	178.54 (160.09–196.83)	216.92 (198.78–235.32)	0.21 (0.18–0.26)	4398.75 (3939.02–4922.37)	5469.26 (5025.00–5994.85)	0.24 (0.21–0.30)	9.38 (7.02–12.05)	7.73 (6.03–9.39)	−0.18 (−0.38–0.17)	511.33 (396.10–629.04)	521.63 (396.92–665.45)	0.02 (−0.10− 0.15)
	Democratic People’s Republic of Korea	165.53 (151.06–180.48)	212.41 (193.22–234.28)	0.28 (0.22–0.36)	3967.84 (3579.20–4393.02)	5257.18 (4665.80–5847.54)	0.32 (0.25–0.41)	15.27 (9.58–22.54)	15.10 (10.01–22.18)	−0.01 (−0.32–0.62)	650.86 (490.41–826.92)	740.66 (555.56–954.85)	0.14 (−0.07− 0.45)
	High-income Asia Pacific	243.89 (220.64–267.00)	266.50 (247.61–287.32)	0.09 (0.05–0.13)	5237.51 (4719.44–5702.44)	5341.92 (4920.36–5766.25)	0.02 (−0.01− 0.05)	7.28 (6.51–8.11)	4.50 (3.81–5.24)	−0.38 (−0.47–−0.29)	544.49 (426.13–682.24)	519.92 (382.34–668.19)	−0.05 (−0.11− 0.02)
	Brunei Darussalam	295.66 (276.69–317.93)	384.53 (358.09–416.73)	0.30 (0.23–0.38)	5736.79 (5296.14–6240.39)	7974.73 (7356.05–8624.55)	0.39 (0.31–0.47)	74.90 (54.90–94.16)	52.50 (40.57–67.28)	−0.30 (−0.51− 0.03)	2105.43 (1684.33–2555.93)	1775.76 (1448.39–2122.76)	−0.16 (−0.36− 0.12)
	Japan	239.33 (213.83–264.60)	262.39 (237.57–290.63)	0.10 (0.05–0.14)	5228.05 (4675.93–5798.99)	5396.76 (4831.15–5983.29)	0.03 (0.00–0.06)	5.06 (4.42–5.65)	3.27 (2.76–3.74)	−0.35 (−0.46–−0.25)	475.94 (364.69–616.51)	501.66 (369.06–659.88)	0.05 (−0.01− 0.12)
	Republic of Korea	271.78 (251.01–294.08)	284.82 (275.94–294.88)	0.05 (−0.02− 0.12)	5457.77 (5046.63–5899.24)	5397.54 (5231.40–5553.40)	−0.01 (−0.08− 0.05)	19.19 (16.11–24.77)	9.96 (7.65–12.97)	−0.48 (−0.58–−0.38)	877.06 (711.09–1036.48)	602.81 (463.79–761.95)	−0.31 (−0.40–−0.23)
	Singapore	243.23 (224.29–268.41)	193.38 (175.56–215.51)	−0.20 (−0.26–−0.13)	4908.06 (4516.13–5408.78)	4202.06 (3834.33–4698.08)	−0.14 (−0.20–−0.07)	16.21 (14.95–17.35)	2.89 (2.48–3.26)	−0.82 (−0.85–−0.80)	732.92 (612.23–870.79)	380.42 (274.44–517.95)	−0.48 (−0.56–−0.41)
T1DM	East Asia	1.86 (1.55–2.27)	2.64 (1.85–4.21)	0.42 (−0.07− 1.29)	50.63 (37.85–64.90)	61.20 (52.12–71.77)	0.21 (−0.01− 0.53)	0.62 (0.42–0.88)	0.22 (0.16–0.33)	−0.65 (−0.77–−0.51)	30.27 (22.17–40.22)	13.49 (10.48–18.15)	−0.55 (−0.67–−0.43)
	China	1.85 (1.54–2.26)	2.63 (1.84–4.18)	0.42 (−0.07 - 1.28)	50.41 (37.65–64.66)	60.64 (51.62–71.14)	0.20 (−0.01–0.52)	0.59 (0.39–0.85)	0.19 (0.13–0.29)	−0.69 (−0.80–−0.55)	29.35 (20.97–39.64)	12.25 (9.21–16.50)	−0.58 (−0.70–−0.46)
	Democratic People’s Republic of Korea	1.90 (1.60–2.30)	2.88 (1.97–5.05)	0.52 (0.00–1.65)	50.13 (37.87–63.59)	68.52 (58.67–80.33)	0.37 (0.11–0.74)	0.88 (0.41–1.74)	0.67 (0.32–1.44)	−0.24 (−0.55− 0.28)	40.05 (22.63–67.44)	33.71 (20.10–61.18)	−0.16 (−0.45− 0.25)
	High-income Asia Pacific	1.41 (1.18–1.70)	2.58 (1.64–5.46)	0.84 (0.09–2.88)	33.05 (24.76–42.02)	50.09 (42.28–61.82)	0.52 (0.20–1.10)	0.43 (0.35–0.53)	0.16 (0.13–0.21)	−0.63 (−0.71–−0.48)	22.17 (18.24–26.21)	10.23 (8.45–12.88)	−0.54 (−0.62–−0.41)
	Brunei Darussalam	1.57 (1.33–1.89)	2.55 (1.68–4.86)	0.63 (0.06–2.08)	34.04 (25.34–43.49)	51.97 (43.73–63.41)	0.53 (0.20–1.07)	3.18 (1.97–4.87)	1.49 (0.94–2.48)	−0.53 (−0.73–−0.18)	141.08 (91.60–213.26)	70.36 (46.45–110.69)	−0.50 (−0.70–−0.16)
	Japan	1.31 (1.10–1.60)	1.85 (1.27–3.00)	0.41 (−0.09− 1.33)	31.51 (23.70–40.21)	41.93 (35.67–50.51)	0.33 (0.06–0.76)	0.27 (0.23–0.31)	0.15 (0.12–0.17)	−0.46 (−0.57–−0.33)	14.46 (12.68–16.49)	9.21 (7.75–11.03)	−0.36 (−0.47–−0.24)
	Republic of Korea	1.63 (1.37–1.96)	4.33 (2.36–11.43)	1.66 (0.42–6.00)	36.76 (27.08–48.17)	67.46 (55.23–96.93)	0.84 (0.41–1.86)	1.00 (0.67–1.46)	0.19 (0.12–0.36)	−0.81 (−0.87–−0.64)	45.45 (31.49–63.20)	12.37 (9.15–18.98)	−0.73 (−0.80–−0.55)
	Singapore	1.56 (1.33–1.88)	2.54 (1.67–4.84)	0.63 (0.06–2.08)	34.36 (25.65–43.94)	52.57 (44.42–64.18)	0.53 (0.21–1.06)	0.26 (0.23–0.30)	0.07 (0.06–0.08)	−0.73 (−0.79–−0.67)	13.64 (11.79–15.65)	6.58 (5.23–8.24)	−0.52 (−0.61–−0.40)
T2DM	East Asia	177.69 (159.54–195.89)	215.07 (196.84–233.76)	0.21 (0.18–0.25)	4354.97 (3898.54–4868.86)	5416.71 (4968.69–5950.43)	0.24 (0.21–0.29)	9.35 (7.14–12.02)	8.04 (6.34–9.78)	−0.14 (−0.35− 0.19)	496.44 (387.88–604.14)	521.91 (398.95–662.31)	0.05 (−0.06− 0.17)
	China	176.69 (158.22–195.12)	214.30 (196.11–233.14)	0.21 (0.18–0.25)	4348.34 (3885.84–4867.67)	5408.63 (4958.53–5944.63)	0.24 (0.21–0.30)	8.79 (6.53–11.42)	7.54 (5.83–9.21)	−0.14 (−0.36–0.23)	481.97 (373.34–588.73)	509.38 (386.79–650.05)	0.06 (−0.06− 0.18)
	Democratic People’s Republic of Korea	163.64 (149.40–178.63)	209.53 (189.99–231.51)	0.28 (0.22–0.35)	3917.70 (3532.12–4344.48)	5188.66 (4594.77–5783.98)	0.32 (0.25–0.41)	14.40 (9.07–21.05)	14.43 (9.63–21.04)	0.00 (−0.32–0.65)	610.82 (465.78–776.47)	706.95 (528.91–899.77)	0.16 (−0.05− 0.46)
	High-income Asia Pacific	242.48 (219.24–265.75)	263.91 (244.87–284.76)	0.09 (0.05–0.13)	5204.47 (4681.74–5671.92)	5291.84 (4876.58–5719.20)	0.02 (−0.01− 0.05)	6.85 (6.09–7.64)	4.34 (3.66–5.07)	−0.37 (−0.46–−0.27)	522.32 (406.62–658.37)	509.68 (373.83–657.34)	−0.02 (−0.09− 0.04)
	Brunei Darussalam	294.09 (275.21–316.25)	381.98 (355.92–414.70)	0.30 (0.23–0.38)	5702.75 (5256.58–6207.05)	7922.76 (7299.65–8573.13)	0.39 (0.31–0.47)	71.72 (52.62–90.13)	51.02 (39.61–65.31)	−0.29 (−0.50–0.02)	1964.35 (1581.04–2404.01)	1705.40 (1390.20–2024.19)	−0.13 (−0.33− 0.14)
	Japan	238.02 (212.68–263.31)	260.54 (235.35–288.69)	0.09 (0.05–0.13)	5196.55 (4637.65–5767.95)	5354.83 (4794.35–5942.46)	0.03 (0.00–0.06)	4.79 (4.18–5.35)	3.12 (2.62–3.58)	−0.35 (−0.45–−0.24)	461.49 (350.96–601.10)	492.44 (360.64–649.64)	0.07 (0.00–0.13)
	Republic of Korea	270.16 (249.42–292.32)	280.49 (271.94–289.48)	0.04 (−0.03− 0.11)	5421.01 (5011.24–5859.32)	5330.08 (5165.25–5491.66)	−0.02 (−0.08–0.05)	18.19 (15.14–23.62)	9.76 (7.51–12.65)	−0.46 (−0.56–−0.35)	831.61 (674.69–987.79)	590.44 (452.50–749.90)	−0.29 (−0.37–−0.21)
	Singapore	241.67 (222.83–266.93)	190.84 (173.45–213.44)	−0.21 (−0.27–−0.14)	4873.70 (4480.89–5380.39)	4149.49 (3783.15–4643.18)	−0.15 (−0.21–−0.08)	15.95 (14.72–17.07)	2.82 (2.41–3.19)	−0.82 (−0.85–−0.80)	719.29 (599.52–854.95)	373.83 (268.44–509.83)	−0.48 (−0.56–−0.40)

Data are presented as point estimates of age-standardized rates, with 95% UI shown in parentheses.

At the national level, China and the Democratic People’s Republic of Korea in East Asia exhibited different patterns: the two countries had similar ASIR (216.92 vs. 212.41 per 100,000) and ASPR (5469.26 vs. 5257.18 per 100,000), but the Democratic People’s Republic of Korea had significantly higher ASMR (15.10 vs. 7.73 per 100,000) and ASDR (740.66 vs. 521.63 per 100,000) than China. In High-income Asia Pacific, Brunei Darussalam had the highest values for all four burden indicators, followed by Republic of Korea, Japan, and Singapore: ASIR were 384.53, 284.82, 262.39, and 193.38 per 100,000, respectively; ASPR were 7974.73, 5397.54, 5396.76, and 4202.06 per 100,000, respectively; ASMR were 52.50, 9.96, 3.27, and 2.89 per 100,000, respectively; ASDR were 1775.76, 602.81, 501.66, and 380.42 per 100,000, respectively (Table 1; Figure 1).

Figure 1.

Age-standardized disease burden of DM in East Asia and High-income Asia Pacific in 2023

(A) ASIR.

(B) ASPR.

(C) ASMR.

(D) ASDR. Data are presented as point estimates of age-standardized rates, represented by bar length. Maps use red markers to indicate the highest burden and blue markers to indicate the lowest burden; the left panel shows geographic distribution visualization maps, and the right panel presents specific values for each region in bar chart format.

Type 1 diabetes: low absolute rates but up to 20-fold differences in mortality across countries

In 2023, the absolute burden level of T1DM in East Asia and High-income Asia Pacific was substantially lower than that of DM. Specifically, in East Asia, ASIR was 2.64 per 100,000 (95% UI: 1.85–4.21), ASPR was 61.20 per 100,000 (95% UI: 52.12–71.77), ASMR was 0.22 per 100,000 (95% UI: 0.16–0.33), and ASDR was 13.49 per 100,000 (95% UI: 10.48–18.15). In High-income Asia Pacific, ASIR was 2.58 per 100,000 (95% UI: 1.64–5.46), ASPR was 50.09 per 100,000 (95% UI: 42.28–61.82), ASMR was 0.16 per 100,000 (95% UI: 0.13–0.21), and ASDR was 10.23 per 100,000 (95% UI: 8.45–12.88).

At the national level, China and the Democratic People’s Republic of Korea had similar ASIR (2.63 vs. 2.88 per 100,000), but the Democratic People’s Republic of Korea’s ASPR (68.52 per 100,000), ASMR (0.67 per 100,000), and ASDR (33.71 per 100,000) were all higher than China’s. The ranking of burden indicators among the four High-income Asia Pacific countries was inconsistent across dimensions: the Republic of Korea had the highest ASIR (4.33 per 100,000) and ASPR (67.46 per 100,000), while Japan had the lowest (ASIR 1.85 per 100,000, ASPR 41.93 per 100,000); Brunei Darussalam had the highest ASMR (1.49 per 100,000) and ASDR (70.36 per 100,000), while Singapore had the lowest (ASMR 0.07 per 100,000, ASDR 6.58 per 100,000) (Table 1).

Type 2 diabetes: accounts for >94% of cases and dominates the overall DM burden pattern

T2DM constituted the absolute dominant component of DM disease burden, accounting for over 94% of DM cases in both East Asia and High-income Asia Pacific. In East Asia, the ASIR of T2DM was 215.07 per 100,000 (95% UI: 196.84–233.76), the ASPR was 5416.71 per 100,000 (95% UI: 4968.69–5950.43), the ASMR was 8.04 per 100,000 (95% UI: 6.34–9.78), and the ASDR was 521.91 per 100,000 (95% UI: 398.95–662.31). In High-income Asia Pacific, the ASIR was 263.91 per 100,000 (95% UI: 244.87–284.76), the ASPR was 5291.84 per 100,000 (95% UI: 4876.58–5719.20), the ASMR was 4.34 per 100,000 (95% UI: 3.66–5.07), and the ASDR was 509.68 per 100,000 (95% UI: 373.83–657.34).

The national distribution pattern of T2DM was almost identical to that of DM. In East Asia, China and the Democratic People’s Republic of Korea exhibited distinct epidemiological patterns. The two regions had similar ASIR (214.30 vs. 209.53 per 100,000) and ASPR (5408.63 vs. 5188.66 per 100,000), whereas the Democratic People’s Republic of Korea presented markedly higher ASMR (14.43 vs. 7.54 per 100,000) and ASDR (706.95 vs. 509.38 per 100,000) than China. In High-income Asia Pacific, Brunei Darussalam recorded the highest values across all four burden indicators, followed by the Republic of Korea and Japan, while Singapore had the lowest estimates. The corresponding ASIR values were 381.98, 280.49, 260.54, and 190.84 per 100,000; ASPR were 7922.76, 5354.83, 5330.08, and 4149.49 per 100,000; ASMR were 51.02, 9.76, 3.12, and 2.82 per 100,000; ASDR were 1705.40, 590.44, 492.44, and 373.83 per 100,000, respectively (Table 1).

Mortality-to-incidence ratio reveals profound inequalities in diabetes care between and within regions

In 2023, the MIR of DM exhibited marked regional disparities and internal heterogeneity across East Asia and the High-income Asia Pacific. The MIR has been widely adopted to evaluate cross-national quality of diabetes care, with lower MIR values typically indicating timely diagnosis and effective clinical management for affected patients.

DM MIR: East Asia’s 0.04 doubles the 0.02 in the high-income Asia Pacific; Brunei Darussalam reaches 0.14, the highest in either region

At the regional level, the DM MIR was 0.02 (95% UI: 0.01–0.02) in High-income Asia Pacific, whereas the figure reached 0.04 (95% UI: 0.03–0.05) in East Asia, suggesting a substantially higher relative mortality risk among patients with DM in East Asia. At the national level, inter-country discrepancies within the two regions were even more prominent. In High-income Asia Pacific, Japan and Singapore shared the lowest MIR at 0.01 (both 95% UI: 0.01–0.02); the Republic of Korea ranked intermediate with an MIR of 0.03 (95% UI: 0.03–0.05); Brunei Darussalam recorded an extremely high MIR of 0.14 (95% UI: 0.10–0.19), considerably higher than that of other regional counterparts. In East Asia, China’s MIR was 0.04 (95% UI: 0.03–0.05), consistent with the regional average; the Democratic People’s Republic of Korea presented an MIR of 0.07 (95% UI: 0.04–0.11), which was significantly higher than that of China (Figure 2A).

Figure 2.

MIR of DM in East Asia and High-income Asia Pacific in 2023

(A) DM.

(B) T1DM.

(C) T2DM. Green represents DM, pink represents T1DM, and blue represents T2DM. Data are presented as point estimates (bar height) for the age-standardized MIR, with 95% UI shown as error bars.

T1DM MIR: substantially higher than T2DM MIR, peaking at 0.58 in Brunei Darussalam

The MIR of T1DM was markedly higher than that of DM, reflecting a higher mortality vulnerability of T1DM among children, adolescents, and young adults. At the regional level within East Asia, China had a T1DM MIR of 0.07 (95% UI: 0.03–0.16), while the Democratic People’s Republic of Korea reached 0.23 (95% UI: 0.06–0.73), approximately three times the estimate for China. Regionally, the T1DM MIR was 0.06 (95% UI: 0.02–0.13) in High-income Asia Pacific and 0.08 (95% UI: 0.04–0.18) in East Asia, with a slightly higher level in East Asia. Nevertheless, the national-level distribution pattern across regions was more complex.

In High-income Asia Pacific, Singapore had the lowest T1DM MIR at 0.03 (95% UI: 0.01–0.05); Republic of Korea and Japan recorded 0.04 (95% UI: 0.01–0.15) and 0.08 (95% UI: 0.04–0.13), respectively; Brunei Darussalam showed an extraordinarily high MIR of 0.58 (95% UI: 0.19–1.48), which far exceeded not only other nations in the region but also all individual countries in East Asia (Figure 2B).

T2DM MIR: patterns mirror total DM, driven entirely by the predominance of T2DM

The T2DM MIR pattern was highly consistent with that of DM, and its distribution trend largely determined the overall epidemiological characteristics of DM. At the regional level, the T2DM MIR was 0.04 (95% UI: 0.03–0.05) in East Asia and 0.02 (95% UI: 0.01–0.02) in High-income Asia Pacific. At the national level, China recorded a T2DM MIR of 0.04 (95% UI: 0.03–0.05) and the Democratic People’s Republic of Korea recorded 0.07 (95% UI: 0.04–0.11) in East Asia. In High-income Asia Pacific, the ranking of Brunei Darussalam, Japan, the Republic of Korea, and Singapore was identical to that for DM: Japan and Singapore tied for the lowest MIR (both 0.01, 95% UI: 0.01–0.02), and Brunei Darussalam had the highest value (0.13, 95% UI: 0.10–0.18). T2DM accounted for the vast majority (>94%) of all DM cases across all countries and regions. Accordingly, the overall MIR trend was primarily driven by T2DM (Figure 2C).

Contrasting 1990–2023 trends: T1DM incidence rises while mortality falls; T2DM burden escalates in East Asia but improves in the high-income Asia Pacific, with recent upturns

Between 1990 and 2023, the disease burden of DM underwent remarkable dynamic evolution in East Asia and High-income Asia Pacific (Figure 3A; Table S3). In-depth analysis indicated that T1DM and T2DM presented distinctly different epidemiological transition patterns. Given the inherent discrepancies in their temporal trends, subsequent analyses focused on the independent evolutionary trajectories of T1DM and T2DM.

Figure 3.

Joinpoint regression analysis

(A) Trends in DM-related incidence, prevalence, mortality, and DALYs in East Asia and High-income Asia Pacific, 1990–2023.

(B) Trends in T1DM-related incidence, prevalence, mortality, and DALYs in East Asia and High-income Asia Pacific, 1990–2023.

(C) Trends in T2DM-related incidence, prevalence, mortality, and DALYs in East Asia and High-income Asia Pacific, 1990–2023. The vertical axis represents different countries/regions, and the horizontal axis represents the time span from 1990 to 2023. The Joinpoint regression model can identify significant inflection points (p < 0.05) in incidence, prevalence, mortality, and DALYs, and calculate APC for each time period. Data are presented as point estimates of the APC. APC values marked with asterisks (∗) indicate statistically significant trends (p < 0.05). Different colors represent different epidemiological characteristic periods, with joinpoints identifying trend directions (red indicates an increase, blue indicates a decrease).

Trends of T1DM in East Asia and high-income Asia Pacific, 1990–2023

In both East Asia and High-income Asia Pacific, T1DM burden indicators exhibited a unique “two rises and two falls” pattern. Specifically, ASIR and ASPR of T1DM showed overall upward trends in East Asia, with TPC of 0.42 and 0.21, respectively; ASMR and ASDR declined markedly, with TPC of −0.65 and −0.55. Similarly, in High-income Asia Pacific, ASIR and ASPR increased overall (TPC: 0.84 and 0.52), whereas ASMR and ASDR decreased substantially (TPC: −0.63 and −0.54) (Table 1).

Joinpoint regression analysis revealed that ASIR and ASPR of T1DM maintained consistent upward in both regions from 1990 to 2023. Notably, in High-income Asia Pacific, ASMR and ASDR initially declined after 1990 but have trended upward since 2015; in East Asia, ASMR and ASDR declined continuously throughout the study period. Country-level temporal trends were generally consistent with regional patterns (Figure 3B; Table S4).

Trends of T2DM in East Asia and high-income Asia Pacific, 1990–2023

The overall temporal trend of T2DM was similar to that of T1DM across the two regions, except that only ASDR presented an overall decreasing trend in East Asia. From 1990 to 2023, ASIR, ASPR, and ASDR in East Asia increased by 21%, 24%, and 5%, respectively, while ASMR decreased by 14%. In High-income Asia Pacific, ASIR and ASPR rose by 9% and 2%, whereas ASMR and ASDR dropped by 37% and 2% (Table 1).

Joinpoint regression further identified complex fluctuations beneath the overall trends. In East Asia, all four T2DM burden indicators experienced multiple fluctuating phases and eventually turned downward around 2020. In High-income Asia Pacific, ASIR and ASPR of T2DM fluctuated across stages and finally showed a net increase after 2021; ASMR and ASDR initially increased, followed by a sustained decline of more than 20 years, before shifting to a marked upward trend after 2020.

National trends were largely consistent with regional patterns, with several notable exceptions. For instance, ASIR, ASPR, and ASDR in Singapore all exhibited long-term significant declines. ASIR and ASPR in the Democratic People’s Republic of Korea maintained a sustained upward trend throughout 1990–2023 (Figure 3C; Table S5).

Younger East Asian adults and males bear a disproportionate DALY burden

In 2023, age-specific DALYs in East Asia increased with age and fluctuated slightly after peaking in the 80–84 age group. By contrast, DM-related DALYs in High-income Asia Pacific rose steadily across all age groups. Notably, within 0–49, DALYs in East Asia were consistently higher than those in High-income Asia Pacific. For 50+ years old, DALYs in High-income Asia Pacific exceeded those in East Asia, with the disparity widening progressively with age (Figure 4A).

Figure 4.

Age- and sex-specific DALYs of DM in East Asia and High-income Asia Pacific in 2023

(A) Age-standardized DALYs distribution for DM across age groups in East Asia and High-income Asia Pacific in 2023, with red representing East Asia and blue representing High-income Asia Pacific.

(B) Trends in DM-related DALYs by sex in East Asia, 1990–2023.

(C) Trends in DM-related DALYs by sex in High-income Asia Pacific, 1990–2023. Blue lines represent males, red lines represent females, and orange lines represent both sexes combined; Data are presented as point estimates of age-standardized rates, with 95% UI shown as error bars in (A) and as shaded bands in (B) and (C). The 95% UI represents the range within which the true value lies with 95% probability.

In terms of sex disparities, East Asia recorded higher DALYs in females than in males from 1990 to 1997. After 1997, male DALYs overtook female levels and remained persistently higher thereafter (Figure 4B). In High-income Asia Pacific, DM incidence was consistently higher in males than in females throughout 1990–2023 (Figure 4C).

Major risk factor attribution analysis for DM DALYs

In East Asia and High-income Asia Pacific, BMI was the dominant driver of T2DM among the 17 risk factors attributable to DM-related ASDR, with attributable proportions reaching 49.30% and 48.82% in in 2023, respectively. The ranking of other major risk factors differed between the two regions. In East Asia, the second leading risk factor was ambient particulate matter pollution (16.28%), followed by smoking (11.11%), a diet high in red meat (8.53%), and low physical activity (6.48%). In High-income Asia Pacific, ambient particulate matter pollution (13.53%) also ranked second, followed by a diet high in processed meat (11.86%), smoking (9.62%), and low physical activity (9.42%).

Distinct from the risk attribution pattern of T2DM, T1DM burden was primarily linked to ambient temperature exposure. In 2023, the proportion of T1DM DALYs attributable to low temperature was 4.15% in East Asia and 4.74% in High-income Asia Pacific, while the attributable fraction of high temperature was 0.98% and 0.79%, respectively (Figure 5A; Table S6).

Figure 5.

Attributable proportions of risk factors for T1DM and T2DM DALYs in East Asia and High-income Asia Pacific in 2023

(A) T1DM.

(B) T2DM. Green represents East Asia, and orange represents High-income Asia Pacific. Data are presented as point estimates (bar length) of the attributable proportion.

Between 1990 and 2023, the attributable proportions of key risk factors underwent substantial changes. For T2DM, the contributions of most metabolic and environmental risk factors continued to increase over the period. In East Asia, the attributable fractions of diet high in sugar-sweetened beverages, ambient particulate matter pollution, diet high in processed meat, and high BMI to T2DM DALYs surged by 426.84%, 264.33%, 133.31%, and 37.55%, respectively. By contrast, the attributable burdens of high alcohol use, household air pollution from solid fuels, a diet low in vegetables, and a diet low in fiber decreased by 140.82%, 83.33%, 71.43%, and 52.61%, respectively.

In High-income Asia Pacific, the attributable proportions of diet high in processed meat, diet low in whole grains, diet low in fiber, high BMI, and diet high in sugar-sweetened beverages dietary pattern increased by 71.73%, 25.21%, 22.14%, 20.62%, and 20.52%, respectively. Meanwhile, notable declines were observed for high alcohol use (−92.01%), household air pollution from solid fuels (−87.88%), low-temperature (41.85%), secondhand smoke (−36.84%), and smoking (−28.31%).

For T1DM, temperature-related risk factors exhibited opposite temporal trends. From 1990 to 2023, the attributable fraction of high temperature exposure to T1DM DALYs increased by 15.65% in East Asia and 29.54% in High-income Asia Pacific, whereas the attributable proportion of low-temperature exposure decreased by 32.6% and 30.37%, respectively. This contrasting trend underscores the growing contribution of climate change-related heat stress to T1DM disease risk (Figure 5B; Table S7).

Projections to 2050: continuing rise in DM and T2DM DALYs under current trends, but optimized interventions could avert >84% of DALYs in East Asia and >70% in the high-income Asia Pacific

Based on current epidemiological trends, this study projected the disease burden of DM and its subtypes through 2050. Under the reference scenario that maintains existing temporal trends, the DALY burden of DM is expected to climb continuously in both East Asia and High-income Asia Pacific.

By 2050, the overall ASDR in East Asia is projected to reach 1720.67 per 100,000 (95% UI: 1285.51–2246.47). Country-specific estimates show an ASDR of 1698.54 per 100,000 (95% UI: 1260.69–2228.45) for China and 1435.91 per 100,000 (95% UI: 1095.78–1873.76) for the Democratic People’s Republic of Korea. In High-income Asia Pacific, the overall ASDR will rise to 1834.73 per 100,000 (95% UI: 1349.43–2460.17). Brunei Darussalam is projected to have the highest burden at 4058.87 per 100,000 (95% UI: 2987.00–5249.05), followed by Republic of Korea (2732.45 per 100,000, 95% UI: 2020.77–3641.27), Singapore (2003.37 per 100,000, 95% UI: 1381.29–2840.59), and Japan (1421.80 per 100,000, 95% UI: 1018.54–1952.42) (Figure 6A; Table S8).

Figure 6.

Projections of ASDR for DM in East Asia and High-income Asia Pacific under five different scenarios in 2050

(A) Projected trends in DM ASDR for East Asia and High-income Asia Pacific from 2022 to 2050 under the reference scenario.

(B) Projected ASDR for DM in East Asia and High-income Asia Pacific in 2050 under four intervention scenarios. Different colors represent different regions or countries, with shaded areas corresponding to each color representing 95% UI, reflecting the statistical uncertainty of the projections. Data are presented as point estimates of age-standardized rates (lines) with 95% UI shown as shaded bands.

Stratified by subtype, T2DM serves as the primary driver of the projected burden growth. The ASDR of T2DM is predicted to increase to 1704.93 per 100,000 in East Asia and 1806.61 per 100,000 in High-income Asia Pacific, whereas the burden of T1DM is expected to maintain a sustained declining trend (Tables S9, S10, and S11).

Nevertheless, projection results suggest that such a severe future burden can be reversed through targeted public health interventions. Under the optimized behavioral and metabolic intervention scenario, effective population-level measures targeting major modifiable risk factors (e.g., high BMI, physical inactivity, and unhealthy dietary patterns) could lead to a substantial reduction in the burden of all DM subtypes.

Under this ideal intervention scenario.

• (1)The overall DM ASDR in East Asia could be reduced to 275.82 per 100,000 (95% UI: 193.48–386.32), a decline of over 84% relative to the reference scenario; the corresponding ASDR in High-income Asia Pacific could decline to 543.21 per 100,000 (95% UI: 387.08–746.87), representing a reduction of more than 70%.
• (2)The T1DM ASDR could drop to 10.97 per 100,000 (95% UI: 7.56–15.57) in East Asia and 25.91 per 100,000 (95% UI: 16.78–36.96) in High-income Asia Pacific.
• (3)The T2DM ASDR could decrease to 264.84 per 100,000 (95% UI: 185.22–371.95) in East Asia and 517.30 per 100,000 (95% UI: 367.18–715.08) in High-income Asia Pacific.

Moreover, all individual countries across the two regions are projected to achieve considerable declines in ASDR under this optimal scenario (Figure 6B; Tables S9, S10, and S11). These findings clearly demonstrate that interventions targeting behavioral and metabolic risk factors yield far greater public health benefits than alternative strategies and constitute the fundamental approach for long-term DM prevention and control. Nonetheless, large-scale and sustained population-level behavioral modification remains confronted with substantial practical challenges, including low public compliance, limited health resource investment, and inherent sociocultural disparities.

Discussion

Based on the GBD 2023, this study systematically illustrated the temporal evolution and complex epidemiological landscape of DM burden in East Asia and High-income Asia Pacific from 1990 to 2023. Our findings confirm that DM remains a pressing global public health emergency, with its overall burden continuing to rise as previously reported.¹⁴

True determinants of DM outcomes

This study identified substantial geographical heterogeneity in DM burden across East Asia and High-income Asia Pacific, which was predominantly driven by T2DM. The remarkable cross-country disparities in burden patterns, especially the striking gap in MIR, reflect structural inequities in healthcare resource allocation and capacity for full-cycle clinical management of DM.¹⁵

Regionally, East Asia presented a considerably higher DM mortality rate than the High-income Asia Pacific (8.26 vs. 4.50 per 100,000), whereas its incidence rate was markedly lower. Consequently, the MIR in East Asia was twice that of the High-income Asia Pacific (0.04 vs. 0.02). Even among high-income economies, inter-country discrepancies remained prominent. The four high-income nations differed in DM incidence, with even more pronounced hierarchical disparities in MIR, ranging from 0.01 to 0.14.

Notably, Brunei Darussalam, a high-income country in the High-income Asia Pacific, recorded an extremely high MIR of 0.14. This value was far higher than that of China (0.04), a less economically developed setting, and also significantly exceeded that of the Democratic People’s Republic of Korea (0.07), which has long been challenged by a constrained healthcare system capacity. This phenomenon strongly indicates that economic development level alone cannot fully determine DM prognosis. Instead, accessible high-quality diagnosis, standardized treatment, and sustainable long-term disease management serve as the core determinants of DM survival outcomes.^16,17 Likewise, the exceedingly low MIR (0.01) in Japan and Singapore is likely attributable to their well-established primary healthcare systems and optimized chronic disease management frameworks.¹⁸ In addition, this study revealed that T1DM had a remarkably higher MIR (peaking at 0.58) compared with T2DM. This finding aligns with the global challenge of T1DM management, whereby T1DM mortality varies substantially across low- and high-income settings and among individual nations.¹⁹ Although China’s overall MIR remained at a moderate level, its enormous diabetic population imposes a substantial clinical and healthcare burden.

Therefore, in population-level DM prevention and control strategies, reducing premature mortality and disability burden should be prioritized over mere incidence control.²⁰ Establishing an integrated care framework covering early diagnosis, standardized treatment, and lifelong complication management is essential to mitigate regional health disparities.²¹

Dynamic evolution of DM burden: divergent trajectories of T1DM and T2DM

From 1990 to 2023, the temporal evolution of DM exhibited distinct subtype divergence and regional disparities across East Asia and High-income Asia Pacific. T1DM followed a unique “two rises and two falls” epidemiological pattern: incidence and prevalence increased steadily, whereas mortality and DALY rates declined markedly. This trend indicates that advances in clinical diagnosis and treatment have improved patient survival, while the incidence of new cases continues to grow.²² Notably, the ASMR in High-income Asia Pacific has rebounded since 2015, which may be attributed to population aging and growing challenges in long-term complication management. By contrast, the sustained decline of ASMR in East Asia reflects remarkable progress in DM mortality prevention and control in countries including China.²³ Most of the four major burden indicators of T2DM increased across East Asia, highlighting the escalating severity of its diabetes epidemic. These findings are consistent with GBD evidence of a rapid expansion in absolute T2DM case numbers in East Asia, particularly in China,^24,25 and support the viewpoint that Asia is experiencing the most substantial and accelerated rise in T2DM burden worldwide.²⁶ Cross-national heterogeneity was further observed: the ASIR and ASPR maintained continuous growth in the Democratic People’s Republic of Korea, while multiple burden indicators declined steadily in Singapore. Such differences further confirm the critical moderating effects of socioeconomic development and health policy. The Human Development Index (HDI) is positively correlated with T2DM burden across Asian countries,²⁷ while nations with a high socio-demographic index (SDI) can effectively mitigate the DM epidemic through intensified population screening and integrated chronic disease management.²⁸ Similarly, although Pacific island regions do not have the highest global DM incidence, their fragile healthcare infrastructure and limited medical resources contribute to an excessively high and long-standing DM mortality burden that remains poorly ameliorated.²⁹ Furthermore, existing evidence has reported higher suicide mortality and suicidal ideation rates among patients with T1DM compared with those with T2DM,³⁰ emphasizing the necessity of developing subtype-specific strategies for psychological intervention and complication care.

The pronounced internal heterogeneity across regions implies that conventional one-size-fits-all DM policies are insufficient. Prevention and control strategies should differentiate the independent epidemiological features of T1DM and T2DM, and be dynamically optimized according to national socioeconomic conditions, healthcare system capacity, and local risk factor profiles. Health system strengthening remains an indispensable foundation for curbing the DM epidemic.

“Age Shift” in disease burden and Identification of high-risk groups

In terms of age distribution, DALYs among the population aged 0–49 years in East Asia were consistently higher than those in High-income Asia Pacific, indicating a heavier diabetes burden in younger adults. This phenomenon may be driven by the rapidly rising prevalence of overweight and obesity, unhealthy dietary patterns (e.g., high intake of sugar-sweetened beverages and low whole grain consumption), and increasing exposure to ambient particulate pollution among Asian young populations.^26,31,32 For populations aged 50 years and older, DALYs in High-income Asia Pacific surpassed those in East Asia, with the inter-regional gap gradually widening over time. This pattern reflects the compound long-term impacts of population aging and cumulative metabolic risk factors.^11,25 In terms of gender disparity, male DALYs in East Asia overtook female levels after 1997, while males in High-income Asia Pacific consistently sustained a higher disease burden. Such gender differences warrant targeted attention to disparities in smoking prevalence and physical inactivity between males and females.³³

DM risk attribution: BMI Dominance and the Urgency of air pollution control

BMI was the predominant driving factor of T2DM, responsible for 49.3% and 48.82% of T2DM-related DALYs in East Asia and High-income Asia Pacific in 2023, respectively. This attributable proportion has increased consistently since 1990, which aligns well with the global worsening obesity epidemic.¹⁴ Notably, ambient particulate matter pollution stood as the second leading modifiable risk factor, accounting for 16.28% of T2DM DALYs in East Asia and 13.53% in High-income Asia Pacific. Its attributable fraction in East Asia has surged by 264.33% over the past three decades, highlighting an urgent need to integrate air pollution governance into diabetes prevention strategies.^34,35,36 This finding is biologically plausible and consistent with existing evidence demonstrating that PM2.5 exposure is linked to elevated fasting glucose, insulin resistance, and an increased risk of diabetes onset.^37,38,39 Behavioral risk factors, including a diet high in processed meat and low physical activity, have displayed upward trends, whereas traditional risks such as household air pollution from solid fuels and smoking have gradually declined. This shifting pattern reflects the transformation of dominant risk profiles alongside socioeconomic development.^40,41

Of particular concern is the changing risk attribution pattern for T1DM. From 1990 to 2023, the attributable burden of high-temperature exposure on T1DM DALYs increased by 15.65% in East Asia and 29.54% in High-income Asia Pacific, while the contribution of low-temperature exposure decreased by 32.6% and 30.37%, respectively. Such divergent trends are highly suggestive of the potential impacts of global climate change. Extreme weather events may disturb physiological thermoregulation, aggravate systemic inflammation, and further impair metabolic homeostasis among patients with T1DM.^42,43 Nevertheless, these ecological associations remain speculative. Confirming causal mechanisms requires further individual-level cohort studies and laboratory mechanistic validation. Additionally, impaired skin barrier function and delayed wound healing among patients with diabetes deserve greater clinical attention; skin microbiome dysbiosis may accelerate the progression of diabetic foot ulcers and related complications.⁴⁴

Comprehensive prevention and control Implications

Based on the present findings, several targeted public health recommendations are proposed. First, air pollution mitigation should be formally incorporated into integrated diabetes prevention and control frameworks. Strengthened PM2.5 monitoring, emission reduction, and environmental governance are particularly warranted in high-burden areas across East Asia.^32,45 Second, early lifestyle and BMI interventions should prioritize children and young adults, promoting whole-grain and nut consumption while restricting excessive intake of sugar-sweetened beverages and processed meats.^46,47 Third, gender-tailored health education programs should be developed, with intensified behavioral interventions for males regarding smoking cessation, dietary optimization, and regular physical activity.¹³ Furthermore, despite the declining attributable burden of low-temperature exposure for T1DM, the continuous rise in high-temperature-related attribution underscores the necessity of embedding climate change adaptation into long-term chronic disease management. Multi-sectoral collaboration, together with risk-stratified population interventions, represents a fundamental pathway to alleviate the growing regional burden of diabetes.

Overall, DM and its major subtypes constitute significant public health challenges in East Asia and High-income Asia Pacific, with marked regional and national heterogeneity. T2DM is the absolute dominant component of DM, accounting for over 94% of all cases, while the burden of T1DM is significantly lower. The DM-related mortality and health loss burden in East Asia is heavier than that in High-income Asia Pacific. Brunei Darussalam in High-income Asia Pacific and Democratic People’s Republic of Korea in East Asia both exhibit high levels across multiple burden indicators. There are clear differences in MIR of DM and its subtypes between the two regions, suggesting gaps in the quality of DM diagnosis, treatment, and management among different countries. From 1990 to 2023, T1DM and T2DM in the two regions showed distinct epidemiological evolutionary trends, with T2DM being the primary driver of changes in DM burden. Age and sex significantly impact DM burden: high BMI and ambient particulate matter pollution are the main attributable risk factors for T2DM-related DALYs, while T1DM burden is primarily associated with temperature exposure, and the impact of high-temperature exposure is showing an upward trend.

Limitations of the study

This study has several limitations. First, GBD estimates rely on models and existing data, and in regions with weak health systems, may be systematic underestimation of the true burden. Second, this study did not include gestational diabetes, as it is classified under maternal disorders rather than DM in GBD. Given that a history of gestational diabetes significantly increases the long-term risk of T2DM, the T2DM burden estimated in this study may be somewhat underestimated. Nevertheless, through its standardized, peer-reviewed modeling framework, GBD maximally integrates and calibrates available data globally, and its results remain the optimal benchmark for current global and regional comparisons.

Resource availability

Lead contact

Further information and resource requests should be addressed to and fulfilled by the lead contact, Wenping Gong (gwp891015@whu.edu.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •Data: Data reported in this paper are publicly available from the Global Health Data Exchange (GHDx) at https://ghdx.healthdata.org/gbd-2023. Accession codes are provided in the Key Resources Table. Data reported in this paper will be shared by the lead contact upon request.
• •Code: All original code has been deposited at GitHub and is publicly available at https://github.com/JivonKiang/Diabetes_GBD_2023_Asia as of the date of publication. This paper does not report original code.
• •Other items: Any additional information required to reanalyze the data reported in this paper are available from the lead contact upon request.

Acknowledgments

We sincerely thank the reviewers and editors who provided review and editing services for this study. We acknowledge the Institute for Health Metrics and Evaluation (IHME) for providing publicly accessible data from the 2023 Global Burden of Disease Study, which enabled this research. We also sincerely recognize the tremendous efforts of the Global Burden of Disease (GBD) collaborators in generating, validating, and maintaining these estimates, without which this analysis would not have been possible.

Author contributions

Conceptualization, W.G., L.W., and L.Z.; methodology, Y.L., F.J., Y.L., W.S., R.N., Y.A., S.Z., M.Z., and Y.T.; software, F.J.; visualization, Y.L. and F.J.; supervision, W.G.; project administration, W.G.; writing – original draft preparation, Y.L. and F.J.; Writing –review and editing, W.G., L.W., and L.Z.; All authors have read and agreed to the published version of the manuscript.

Declaration of interests

All of the authors declare no conflict of interest.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work, the authors used the DeepSeek (https://chat.deepseek.com/) online webpage in order to improve language only. After using this tool or service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Software and algorithms
R software	R software	Version 4.5.0; https://www.r-project.org/
Joinpoint Regression Program	Joinpoint Regression Program	https://surveillance.cancer.gov/joinpoint/
GBD Analytical Models (CODEm, DisMod-MR 2.1, DisMod-AT, ST-GPR)	GBD Analytical Models (CODEm, DisMod-MR 2.1, DisMod-AT, ST-GPR)	Part of GBD 2023 framework
Other
Global Health Data Exchange (GHDx) 2023	Institute for Health Metrics and Evaluation (IHME)	Accession code: 594431, 594430; https://doi.org/10.1016/S0140-6736(25)01637-X; URL: https://ghdx.healthdata.org/gbd-2023
Analysis code for this study	GitHub	https://github.com/JivonKiang/Diabetes_GBD_2023_Asia

Experimental model and study participant details

This study didn’t use any experimental model.

Method details

Overview

This study followed GBD analytical framework. GBD 2023 represents a global collaborative effort covering 204 countries and territories, involving 375 diseases, injuries, and 88 risk factors. The present study focused on the disease burden of DM and its two major subtypes, T1DM and T2DM, in East Asia and the High-income Asia Pacific from 1990 to 2023.In accordance with the official GBD regional classification, the High-income Asia Pacific consists of four economies including Brunei Darussalam, Japan, Republic of Korea, and Singapore, while East Asia comprises China and the Democratic People’s Republic of Korea.

Methodologically, GBD provided comprehensive disease burden estimation through systematic data integration and standardized modeling procedures. Mortality estimates incorporated vital registration and population surveillance data, with the Cause of Death Ensemble model adopted to quantify cause-specific mortality. GBD 2023 further integrated COVID-19-related covariates to capture the impact of public health emergencies, and years of life lost (YLL) were calculated to quantify premature mortality. For non-fatal burden estimation, multi-source data were retrieved from systematic reviews, population surveys, and medical insurance claim records. The DisMod-MR 2.1 framework was applied to achieve internally consistent modeling of incidence and prevalence, generating age-, sex-, region-, and year-specific estimates. Years lived with disability (YLD) were further calculated using standard disability weights. Beyond the conventional DisMod-MR 2.1, the DisMod-AT model, which allows for the incorporation of age-cohort effects, was specifically adopted for T1DM to better capture rapidly evolving epidemiological patterns. Spatiotemporal Gaussian process regression was employed to smooth sparse longitudinal and geographical data. As the sum of YLL and YLD, DALYs comprehensively quantified total health loss attributable to premature death and chronic disability, serving as the core summary metric for disease burden evaluation in GBD studies.^48,49

Disease Definition

In the GBD study, DM is defined as individuals with a fasting plasma glucose concentration ≥7 mmol/L (126 mg/dL) or those currently using insulin or hypoglycemic medications.¹⁴ The database classifies DM into only T1DM and T2DM⁵⁰; gestational diabetes is categorized under the maternal disorders module. This study used International Classification of Diseases 10 (ICD-10) and ICD-9 to define DM (Table S1).

Data sources

ASIR, ASPR, ASMR, ASDR, and their 95% uncertainty intervals (UIs) for High-income Asia Pacific, East Asia, and their constituent countries were obtained from the Global Health Data Exchange 2023 (GHDx 2023; access link: https://vizhub.healthdata.org/gbd-results/).

Disease burden statistical analysis and Geospatial visualization

The core analytical indicators included ASIR, ASPR, ASMR, and ASDR. All age-standardized rates were calculated with reference to the GBD 2023 world standard population to reflect relative burden levels. Uncertainty was estimated using Monte Carlo simulation methods based on 250 posterior draws generated by GBD 2023, defining 95% UIs as the 2.5th and 97.5th percentiles of the posterior distribution of each indicator.

To reveal geographic distribution patterns, geographic information system techniques were used to integrate the ASR indicators of DM disease burden with spatial map data (world_map) of East Asia and High-income Asia Pacific, producing spatial visualization maps at both regional and national levels.

Calculation of MIR

This study calculated the all-age and both-sex mortality-to-incidence ratio (MIR) to evaluate disease fatality risk and prognostic outcomes. ASMR and ASIR were extracted from the GBD results tool, with both indicators directly standardized according to the GBD global standard population. MIR was defined as the ratio of ASMR to ASIR (MIR = ASMR/ASIR), ranging from 0 to 1; higher values indicate a higher fatality risk and poorer disease prognosis.⁵¹

Temporal trend analysis

To further quantify temporal changes in disease burden, we calculated the total percentage change (TPC) using the formula: $TPC=\frac{{Estimate}_{2023}-{Estimate}_{1990}}{{Estimate}_{1990}}\times 100\%$ to characterize the overall magnitude of change in ASR for DM and its subtypes between 1990 and 2023. Simultaneously, the Joinpoint regression model was employed to analyze temporal trends in DM and its subtypes in East Asia and the High-income Asia Pacific. This model can precisely identify trend inflection points in time series and is suitable for describing long-term non-linear trends in chronic diseases such as DM.⁵² The model selects optimal joinpoints through permutation tests; considering that setting more than 5 joinpoints in early versions of Joinpoint was computationally infeasible, and to maintain comparability and stability of results across different versions, this study set the maximum number of joinpoints to 5. The annual percentage change (APC) was calculated for each time segment to identify turning points in trends and characteristics of different phases. The statistical significance of APC was assessed using two-sided tests, with statistical significance determined based on 95% confidence interval (CI): if the CI did not include 0, it was considered statistically significant. To visually present Joinpoint regression analysis results, this study further constructed trend heatmaps to visualize the direction, magnitude, and significance of APCs in different time periods.

Age-sex analysis

R software version 4.5.0 was used for data processing and visualization. Parameter settings were as follows: cause selected as DM, region set to East Asia andHigh-income Asia Pacific, sex stratification as male, female, and both sexes combined, age covering all-age groups and25 detailed age groups, indicator as DALYs, time range 1990–2023, with remaining parameters set to default.To clarify regional age-specific differences and sex-stratified long-term trends, grouped bar charts with 95% UI error were plotted to compare DALYs differences across age groups between the two regions; line charts with 95% UI shaded bands were plotted by sex to display temporal trends in disease burden from 1990 to 2023.

Risk factor analysis

GBD 2023 used Bayesian models to integrate exposure data, correct for study bias and heterogeneity, and identify the theoretical minimum risk exposure level (TMREL). Population-attributable fractions were calculated to quantify the proportion of disease burden that could be avoided when exposure to risk factors was reduced to the TMREL. Additionally, the summary exposure value was calculated as a weighted measure of exposure prevalence. Estimates were adjusted for mediation effects to prevent overestimation caused by the indirect influence of risk factors on outcomes through intermediate variables.⁵³

GBD 2023 organized risk factors into a four-level hierarchical structure. Level 1 contained four major categories including environmental/occupational, behavioral, and metabolic risks, with subsequent layers further subdivided. Risk factors incorporated into the DM analysis included Ambient particulate matter pollution, Household air pollution from solid fuels, Smoking, Diet low in fruits, Diet low in vegetables, Diet low in whole grains, Diet high in red meat, Diet high in processed meat, Diet high in sugar-sweetened beverages, Diet low in fiber, Low physical activity, Sexual violence against children, High temperature, Low temperature, Secondhand smoke, High alcohol use, High body-mass index. High fasting plasma glucose was excluded from the final analysis, as GBD 2023 attributed 100% of the diabetes burden to this factor.⁵⁴

Projection

Since the GBD 2023 does not yet have an updated projection module, this study used GBD 2021 data and employed a mixed-effects model to predict the 2050 DM disease burden trends, incorporating core health determinants such as the Socio-Demographic Index (SDI) and comprehensive risk factor exposure levels from the GBD database.⁵⁵The study established one reference projection scenario (Past progress continues) and four intervention scenarios: (1) Safe environment scenario (Exposure to household air pollution, unsafe water, sanitation, and hygiene are eliminated by 2050 and non-optimal temperature and particulate matter pollution follow ssp1-1.9 trajectory); (2) Child nutrition improvement and vaccine coverage scenario (Exposure to child growth failure, iron and vitamin A deficiency, and sub-optimal breastfeeding reach 0 by 2050, and vaccine coverage for DTP3, MCV1, MCV2, HiB, PCV, and Rotavirus reach 100% by 2050); (3) Behavioral and metabolic risk optimization scenario (Exposure to all dietary risk factors as well as high LDL cholesterol, dody mass index, fasting plasma glucose, and systolic blood pressure are eliminated by 2050. Smoking prevalence reaches 0 by 2050); (4) Combined scenario incorporating the synergistic effects of the above three interventions.

Quantification and statistical analysis

All statistical analyses were performed using R version 4.5.0 for data processing and graphical visualization. Joinpoint Trend Analysis Software (version 4.5) was used to construct joinpoint regression models for temporal trend analysis. Disease burden estimation relied on the built-in computational tools of the Global Burden of Disease Study 2023 (GBD 2023), including DisMod-MR 2.1, DisMod-AT, spatiotemporal Gaussian process regression, and Monte Carlo simulation. Projections of disease burden to 2050 were performed using the built-in mixed-effects models from GBD 2021.

For temporal trend analysis, joinpoint regression with permutation tests (maximum of 5 joinpoints) was used to identify statistically significant trend inflections; the APC and its two-sided 95% CI were calculated for each segment, and an APC was considered statistically significant if the 95% CI did not include zero. The 95% UIs for all age-standardized rates (ASIR, ASPR, ASMR, ASDR) were derived from Monte Carlo simulations based on 250 posterior draws, with the 95% UI defined as the 2.5th and 97.5th percentiles of the posterior distribution. The central tendency of each estimate is reported as the mean (posterior mean for raw GBD estimates and arithmetic mean for derived indicators). Dispersion is expressed as 95% UIs for burden estimates and as 95% CIs for APC. The MIR was calculated as the ratio of ASMR to ASIR, and its 95% CI was derived based on the 95% UIs of ASMR and ASIR. Age-sex analyses were descriptive only, visualized using grouped bar charts with 95% UI error bars and line charts with 95% UI shaded bands.

Regarding the exact values of n and what n represents: in joinpoint regression, n = 34 (annual observation points from 1990 to 2023); in age-stratified analyses, n = 25 (all detailed age groups defined by GBD); in regional comparisons, n = 2 (the two super-regions); for country-level visualization, n = 6 (four economies in the High-income Asia Pacific plus two countries in East Asia). In risk factor analysis, n = 17 (the number of risk factors finally included after excluding high fasting plasma glucose). For the projection models based on GBD 2021 data, n corresponds to the number of region-year observations from 1990 to 2021 across the two regions and their four constituent countries, all age groups, and both sexes.

All statistical results are presented in the main text, figures, and tables.

Published: June 29, 2026