BACKGROUND: The mortality of sepsis without direct drugs is high. The association between prediabetes, based on a single fasting glucose (FG), or long-term type 2 diabetes mellitus (T2DM) and sepsis remains unclear. METHODS: Of the adults aged ≥20 years who were included in the National Health Screening Program (NHSP) in 2009, 40% were randomly sampled. After excluding patients with type 1 diabetes mellitus, with missing information, and who were diagnosed with sepsis during the wash-out (between 2001 and the NHSP) or 1-year lag period, a cohort comprised of 3,863,323 examinees. Body mass index (BMI) measurements, FG tests, and self-reported questionnaires on health-related behaviors were conducted. Individual information was followed up until 2020 and censored upon the first occurrence of sepsis or death. The incidence of sepsis was compared using a multivariable regression adjusted for age, sex, income, BMI, smoking, drinking, physical activity levels, and chronic diseases. RESULTS: The cohort was divided into those with normal FG (n=2,675,476), impaired fasting glucose (IFG) (n=890,402, 23.0%), T2DM <
5 years (n=212,391, 5.5%), or T2DM for ≥5 years (n=85,054, 2.2%). The groups with IFG (adjusted hazard ratio [aHR], 1.03
95% confidence interval [CI], 1.01 to 1.05), T2DM <
5 years (aHR, 1.43
95% CI, 1.40 to 1.47), and T2DM for ≥5 years (aHR, 1.82
95% CI, 1.77 to 1.87) exhibited significantly higher incidence of sepsis (P<
0.001), with the greatest risk in patients with T2DM aged <
40 years (aHR, 1.96
95% CI, 1.71 to 2.25). CONCLUSION: Patients with long-standing and young-onset T2DM show a substantially high risk of sepsis, emphasizing the need for infection prevention and vaccination.