Metabolic Indices Predicting Type 2 Diabetes in NAFLD Patients

A secondary analysis of 2,370 NAFLD patients from a prospective Japanese cohort study evaluates the predictive ability of twelve metabolic composite indices for incident type 2 diabetes over a 12-year follow-up. The dataset, authored by Nan’nan Chen and shared under a CC-BY-4.0 license, includes hazard ratios and AUC values for indices like TyG-WC, TyG-WHtR, and VAI. It was last updated on 2026-05-01.

Use Cases

• Comparing predictive performance of metabolic indices like TyG-WC and TyG-WHtR for type 2 diabetes onset based on hazard ratios and AUC values.
• Validating risk stratification cut-offs for clinical use based on the optimal TyG-WHtR cut-off of 4.54 identified in the study.
• Conducting subgroup analyses to assess the consistency of index associations across patient demographics mentioned in the description.
• Building multivariate Cox proportional hazards models for diabetes prediction using the described cohort data.

Strengths

• Includes data from a cohort of 2,370 NAFLD patients, providing a substantial sample size for analysis.
• Reports specific statistical results, including hazard ratios (e.g., HR: 1.80 for TyG-WC) and AUC values (e.g., 0.680 for TyG-WHtR).
• Uses a 12-year prospective study design, enabling longitudinal assessment of diabetes risk.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset is very small (26.9 KB), suggesting it may contain only summary statistics or a manuscript rather than raw patient-level data.

Provenance

Source

Secondary analysis of a prospective Japanese cohort study, shared by author Nan’nan Chen.

Collection Method

Multivariate Cox proportional hazards models, ROC/AUC analysis, and restricted cubic spline regression were used.

Time Range

12-year prospective study period.

Freshness

Last updated 2026-05-01 04:16:25; freshness should be verified.

Geography

Japanese cohort.