Key result
Statistical model choice shifts risk factor identification for incident diabetes despite consistent core clinical predictors.
Cohort (n=913)
Statistical model choice alters risk factor conclusions in IFG cohorts with high competing mortality, suggesting joint interpretation of complementary regression methods is needed.
Objectives Impaired fasting glucose (IFG) is a prediabetic precursor of type 2 diabetes (T2DM). Few long-term cohorts with substantial competing mortality compare risk estimates from logistic, Cox and Fine-Gray regression. This study aimed to examine cross-model discrepancies and demonstrate model-dependent bias in IFG risk inference. Methods We followed 913 Chinese IFG adults recruited in Liuzhou, Guangxi (2004–2008) for up to 20 years until 31 December 2024 (median 17.83 years). Three regression models were used to identify fully adjusted T2DM predictors. Results Fasting glucose, body fat percentage and older age were consistent positive predictors; female sex was persistently protective. Body mass index (BMI) ≥28 kg/m 2 and waist-to-hip ratio showed divergent results across models. Hypertension lacked independent effects in all models, and enrollment year was significant only in Cox regression. Conclusions Statistical model choice alters risk factor conclusions in IFG cohorts with high competing mortality. Joint interpretation of complementary regression methods is recommended.
No takes yet. Share an insight, caveat, or question.
Lu et al. (2026) conducted a cohort in Impaired fasting glucose (n=913). Risk factors (e.g., fasting glucose, body fat percentage, age) was evaluated on Incident type 2 diabetes. Statistical model choice altered risk factor conclusions for incident type 2 diabetes in adults with impaired fasting glucose, though fasting glucose, body fat, and age were consistent predictors.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: