J Diabetes. 2025 Dec;17(12):e70172. doi: 10.1111/1753-0407.70172.
ABSTRACT
AIMS: To examine the association of insulin therapy with in-hospital death and whether this association is mediated by elevated plasma volume among chronic heart failure (CHF) patients with type 2 diabetes mellitus (T2DM).
MATERIALS AND METHODS: A retrospective study combining two electronic medical records: the Medical Information Mart for Intensive Care (MIMIC) database and the Tianjin Heart Failure with Integrated Treatment (TJHFIT) database. Propensity score matching (1:2 ratio) was conducted in CHF-T2DM patients to eliminate the differences in demographics, comorbidity, and the severity of diabetes. Then, conditional logistic regression, restricted cubic spline (RCS) modeling, and mediation analysis were processed.
RESULTS: A total of 7997 CHF-T2DM patients were included, with 6112 from MIMIC (2241 received insulin therapy and 3871 not), and 1885 from TJHFIT (911 received insulin therapy and 974 not). Multivariable conditional logistic regression revealed a significant association between insulin therapy during hospitalization and in-hospital death in both cohorts (MIMIC: OR, 1.37 [95% CI, 1.14-1.63]; TJHFIT: OR, 2.56 [95% CI, 1.53-4.27]). Insulin therapy was associated with a higher likelihood of ΔePVS > 0 (MIMIC: OR, 1.11 [95% CI, 1.02-1.28]; TJHFIT: OR, 1.37 [95% CI, 1.13-1.66]). RCS revealed both ePVS and ΔePVS were associated with in-hospital death in a nonlinear fashion. The ePVS at discharge mediated insulin-associated in-hospital death with a proportion of 12.0% (MIMIC) and 13.2% (TJHFIT).
CONCLUSIONS: Insulin therapy was associated with elevated odds of in-hospital death among CHF-T2DM patients, which was mediated by plasma volume.
TRIAL REGISTRATION: The research has been registered (ChiCTR2300077220).
PMID:41320960 | DOI:10.1111/1753-0407.70172