Skip to content

Statistics — type 2 diabetes

Last curated: 2026-08-30. Conflicting estimates are shown side by side and are not averaged.

Global prevalence and projections

Figure Year Population Method Source
536.6 million; 10.5% 2021 Adults 20–79, all diabetes IDF model using 219 sources Sun 2022, PMID 34879977
783.2 million; 12.2% 2045 projection Adults 20–79 IDF demographic projection Sun 2022, PMID 34879977
529 million (95% UI 500–564); age-standardised 6.1% (5.8–6.5) 2021 All ages, all diabetes GBD Bayesian modelling GBD 2021, PMID 37356446
>1.31 billion (1.22–1.39) 2050 projection All ages GBD regression projection GBD 2021, PMID 37356446
T2D = 96.0% (95.1–96.8) of cases 2021 Global Modelled type decomposition GBD 2021, PMID 37356446
506.0 million prevalent; 23.9 million incident 2021 Global T2D Secondary GBD analysis Huang 2025, PMID 40895618

Geography and disparities

Figure Year/population Method Source
Urban 12.1% vs rural 8.3% 2021 adults 20–79 IDF model PMID 34879977
High-income 11.1% vs low-income 5.5% 2021 adults 20–79 IDF model PMID 34879977
North Africa/Middle East 9.3% (8.7–9.9) age-standardised 2021 GBD model PMID 37356446
Oceania 12.3% (11.5–13.0) age-standardised 2021 GBD model PMID 37356446
Nepal pooled 8.4% (6.2–10.5); urban 8.1%, rural 1.0% Studies 2000–2014 Ten-study meta-analysis Gyawali 2015, PMID 26613684
Slum T2D range 0.9%–25.0% 62 studies Systematic review/meta-analysis Uthman 2022, PMID 35210339

Burden and attributable risk

Figure Year/population Method Source
T2D = 95.4% (94.9–95.9) of diabetes DALYs 2021 global GBD decomposition PMID 37356446
75 million T2D DALYs (63–90) 2021 global GBD secondary analysis Zhang 2024, PMID 39151887
High BMI attributable fraction 52.2% (25.5–71.8) of T2D DALYs 2021 global Comparative risk assessment PMID 37356446
Non-elderly incidence 196.3/100,000 (145.2–257.4) → 361.1 (275.2–458.4) 1990→2021, ages 15–59 GBD analysis He 2025, PMID 40348179
Each one-year older diagnosis: 4% lower mortality, 3% lower macrovascular, 5% lower microvascular risk 1,325,493 adults, 30 countries 26-study meta-analysis adjusted for current age Nanayakkara 2021, PMID 33313987

Costs

Figure Year/population Method Source
US$966 billion 2021 global diabetes health expenditure Attributable-fraction model PMID 34879977
US$1.054 trillion 2045 projection IDF projection PMID 34879977
Direct annual cost US$242–11,917/person Studies 2001–2014 Global systematic review, 2011 international dollars Seuring 2015, PMID 25787932
Indirect annual cost US$45–16,914/person Studies 2001–2014 Same review PMID 25787932
Total annual cost US$29.91–237.38/person in low/lower-middle-income studies Eight studies to 2018 Systematic review Afroz 2018, PMID 30558591

Prevention and remission

Figure Time Population/method Source
DPP lifestyle incidence −58%; metformin −31% Mean 2.8 years 3,234 high-risk adults, RCT Knowler 2002, PMID 11832527
DPPOS lifestyle −27%; metformin −18% 15 years Follow-up by original assignment PMID 26377054
DiRECT remission 46% vs 4% 1 year Cluster RCT, early T2D PMID 29221645
DiRECT remission 36% vs 3% 2 years Trial follow-up PMID 30852132
DiRECT extension remission 13% vs 5% 5 years Extension cohort PMID 38423026
Remission 86% with ≥15 kg loss vs 7% with 0–5 kg 1 year DiRECT subgroup PMID 29221645

Therapy outcomes

Trial statistic Population Source
EMPA-REG MACE HR 0.86; CV death −38%; HF admission −35%; all-cause death −32% T2D + established CVD Zinman 2015, PMID 26378978
LEADER MACE HR 0.87; CV death HR 0.78 T2D high CV risk Marso 2016, PMID 27295427
REWIND MACE 12.0% vs 13.4%; HR 0.88 (0.79–0.99) T2D with/without prior CVD Gerstein 2019, PMID 31189511
CREDENCE kidney composite HR 0.70 T2D + albuminuric CKD Perkovic 2019, PMID 30990260
DAPA-CKD primary composite HR 0.61 CKD with/without diabetes Heerspink 2020, PMID 32970396
SELECT MACE HR 0.80 Obesity + CVD, no diabetes Lincoff 2023, PMID 37952131
Tirzepatide HbA1c change up to −2.30 percentage points T2D, SURPASS-2 Frías 2021, PMID 34170647
Tirzepatide weight −20.9% at 15 mg Obesity without diabetes, 72 weeks Jastreboff 2022, PMID 35658024

Youth-onset disease

Figure Population/time Source
Glycaemic failure 45.6% TODAY, mean 3.9 years PMID 22540912
≥1 microvascular complication 60.1% Mean age 26.4, duration 13.3 years PMID 34320286
≥2 microvascular complications 28.4% Same cohort PMID 34320286

Additional prevalence and care-cascade estimates

Figure Year/population Method Source
828 million adults (95% CrI 757–908 million) 2022; age ≥18; all diabetes Bayesian pooling of 1,108 measured population studies NCD-RisC 2024, PMID 39549716
445 million untreated (401–496 million); 59% 2022; age ≥30 with diabetes Medication coverage in same pooled analysis PMID 39549716
108 million → 422 million 1980→2014; adults; all diabetes 751 population-based studies NCD-RisC 2016, PMID 27061677
28.5% prevalence change; 39.7% population growth; remainder ageing/interaction Contribution to 1980–2014 numerical growth Demographic decomposition PMID 27061677
Diagnosed 55.8% (95% UI 49.3–62.3) 2023; age ≥15 global GBD-linked cascade model Stafford 2025, PMID 40934935
Treated among diagnosed 91.4% (88.0–94.2) Same Same PMID 40934935
Optimal glycaemia among treated 41.6% (35.7–48.5) Same Same PMID 40934935
Optimal glycaemia on treatment among all diabetes 21.2% (17.4–25.6) Same Same PMID 40934935
Comprehensive recommended treatment coverage 4.6% (3.9–5.4) 37,094 adults with diabetes in 55 LMIC surveys Cross-sectional individual-level analysis Flood 2021, PMID 35211689
Glucose/BP/lipid medication coverage 50.5%/41.3%/6.3% Same Self-report, equal country weighting PMID 35211689
US diabetes 14.3%; undiagnosed 5.2% 2011–2012 adults NHANES HbA1c+FPG+OGTT definition Menke 2015, PMID 26348752
US diabetes 12.3%; 25.2% of cases undiagnosed Same NHANES HbA1c+FPG definition without OGTT PMID 26348752

Prevention, remission and surgery deepening

Figure Time/population Method Source
Finnish DPS diabetes 11% vs 23%; relative reduction 58% 4 years; 522 adults with IGT RCT Tuomilehto 2001, PMID 11333990
Incidence 4.3 vs 7.4/100 person-years; 43% relative reduction Median 7 years Post-trial follow-up Lindström 2006, PMID 17098085
IDPP-1 3-year incidence 55.0% control vs 39.3% lifestyle, 40.5% metformin, 39.5% combination 531 Asian Indian adults with IGT RCT Ramachandran 2006, PMID 16391903
DPP 10-year cumulative reduction 34% lifestyle, 18% metformin Median 10 years Randomised assignment follow-up Knowler 2009, PMID 19878986
Da Qing diabetes HR 0.57 (95% CI 0.41–0.81) 20 years after 6-year intervention Clinic-randomised follow-up Li 2008, PMID 18502303
Da Qing severe retinopathy 9.2% vs 16.2%; HR 0.53 (0.29–0.99) 20 years Follow-up Gong 2011, PMID 21046360
Look AHEAD remission 11.5% vs 2.0% at year 1; 7.3% vs 2.0% at year 4 4,503 adults RCT ancillary analysis Gregg 2012, PMID 23288372
Real-world consensus remission 3.2%; stricter definitions 0.8%–2.3% 338,791 assessable insured US adults Claims plus laboratory cohort Sheils 2023, PMID 37583561
Remission-associated CKD HR 0.67 (0.52–0.87), CVD HR 0.60 (0.47–0.79) Look AHEAD follow-up Post-randomisation adjusted association Gregg 2024, PMID 38233592
STAMPEDE HbA1c ≤6.0%: 38% bypass, 24% sleeve, 5% medical 3 years; n=150 RCT Schauer 2014, PMID 24679060
SOS T2D life expectancy +2.1 years (0.2–4.0), mortality HR 0.77 (0.61–0.97) Median 26 years Prospective controlled cohort Carlsson 2023, PMID 37438611

Organ protection and monitoring deepening

Trial/statistic Population/time Source
Steno-2 CV HR 0.47 (0.24–0.73); nephropathy 0.39 (0.17–0.87); retinopathy 0.42 (0.21–0.86) 160 T2D with microalbuminuria; 7.8 years Gaede 2003, PMID 12556541
Steno-2 death HR 0.54 (0.32–0.89) 13.3 years Gaede 2008, PMID 18256393
EMPEROR-Preserved primary 13.8% vs 17.1%; HR 0.79 (0.69–0.90) HF EF >40%, ± diabetes Anker 2021, PMID 34449189
DELIVER primary 16.4% vs 19.5%; HR 0.82 (0.73–0.92) HF EF >40%, ± diabetes Solomon 2022, PMID 36027570
SOLOIST-WHF total-event HR 0.67 (0.52–0.85) T2D after worsening HF Bhatt 2021, PMID 33200892
SOUL MACE 12.0% vs 13.8%; HR 0.86 (0.77–0.96) 9,650 high-risk T2D McGuire 2025, PMID 40162642
SURPASS-CVOT MACE 12.2% vs 13.1%; HR 0.92 (95.3% CI 0.83–1.01) Tirzepatide vs dulaglutide; 13,165 modified ITT Nicholls 2025, PMID 41406444
MOBILE HbA1c difference −0.4 points (−0.8 to −0.1); time-in-range +15 points (8–23) 175 basal-insulin-treated adults; 8 months Martens 2021, PMID 34077499
MASLD prevalence 65.33% (62.35%–68.18%); advanced fibrosis 15.49% in biopsy studies T2D meta-analysis Younossi 2024, PMID 38521116
Charcot foot five-year mortality 24.5% (17.2%–32.6%); amputation 15% Meta-analysis Yammine 2022, PMID 36028441

Youth and acute safety deepening

Figure Population/time Source
Youth T2D incidence 17.9/100,000; annual increase 5.31% (4.46%–6.17%) US ages 10–19, 2017–2018/end of 2002–2018 trend Wagenknecht 2023, PMID 36868256
Youth T2D prevalence 0.34→0.67/1,000; +95.3% (77.0%–115.4%) Six US areas, 2001→2017 Lawrence 2021, PMID 34427600
Liraglutide HbA1c treatment difference −1.06 points Ages 10–<17, 26 weeks Tamborlane 2019, PMID 31034184
Empagliflozin HbA1c difference −0.84 points (−1.50 to −0.19) Ages 10–17, 26 weeks Laffel 2023, PMID 36738751
Combined DKA–HHS mortality OR 2.7 (1.4–4.9) vs isolated crises 1,211 crisis admissions Pasquel 2020, PMID 31704689
HHS inpatient mortality 1.44%→0.77% US admissions 2008→2018 Shaka 2022, PMID 35122906
Severe hypoglycaemia associated death HR 2.69 (1.97–3.67) ADVANCE; 11,140 participants Zoungas 2010, PMID 20925543

Known conflicts and caveats

  • IDF and GBD totals use different age ranges, sources and models; do not average them.
  • “Diabetes” is not always type-specific; T2D is often calculated by modelled subtraction.
  • DALYs and attributable fractions are model outputs with uncertainty wider than sampling error.
  • Cost estimates use incompatible currencies, years, health-system perspectives and included components.
  • Trial relative effects cannot be converted to population impact without baseline risk, adherence and access.
  • Remission denominators differ at five years because extension participation is selective.
  • NCD-RisC, IDF and GBD use different ages, tests and modelling; their totals are parallel estimands.
  • Care-cascade proportions are conditional: treated-among-diagnosed can be high while population control remains low.
  • Post-randomisation remission associations do not prove that remission itself causes lower event rates.
  • Historical surgical cohorts include procedures and medical comparators unlike current practice.
  • Youth projections range widely according to whether incidence is held constant or extrapolated.