Pathophysiological Risk Factors Preceding Incidence of Type 2 Diabetes Subtypes: A Pooled Cohort Study in the United States
Authors' conclusion
Pathophysiological biomarkers were prospectively associated with type 2 diabetes and subtypes, but single-time-point measurements before diagnosis could not reliably distinguish subtypes at diagnosis.
Study context
- Population
- Adults without diabetes from four U.S. cohorts; median follow-up 10 years (range 0–17 years)69,725 observations
- Participants
- 9,661 adults
- Outcome
- Incidence of T2D and subtypes (severe insulin-deficient, severe insulin-resistant, mild obesity-related, mild age-related) by seven pathophysiological biomarkers
- Follow-up
- Median 10 years [range 0-17 years]
Methodological detailsfour U.S. cohorts, time-dependent Cox models, cause-specific hazard ratios (HRs)
- Four U.S. cohorts
- Time-dependent Cox models
- Cause-specific HRs
Main finding
As reported in the abstract
Incident T2D
- Developed T2D
- 1,569
- BMI, HR
- 1.03 (1.02, 1.04) (95% CI)
- SBP per 10 mmHg, HR
- 1.09 (1.05, 1.13) (95% CI)
- HbA1c, HR
- 2.46 (1.73, 3.52) (95% CI)
- HOMA2-%B per 10 units, HR
- 0.89 (0.87, 0.91) (95% CI)
- HOMA2 of insulin resistance (HOMA2-IR), HR
- 1.92 (1.78, 2.07) (95% CI)
- LDL-C per 10 mg/dL, HR
- 1.02 (1.00, 1.04) (95% CI)
- eGFR per 10 mL/min/1.73 m², HR (95% CI); not independently associated
- 1.02 (0.98, 1.07)
Observational data: associations, not effects.
Further findings
Model performance of single-time-point models for subtypes
- C-index
- 0.81-0.90
- F1 scores (modest)
- 0.26-0.51
- Calibration slope (varying sensitivity and calibration)
- 0.22-1.24
Original publication
Pathophysiological Risk Factors Preceding Incidence of Type 2 Diabetes Subtypes: A Pooled Cohort Study in the United States. Diabetes Care. Published 6 February 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →