Whether insulin therapy increases in-hospital mortality in CHF-T2DM remains a key concern in clinical practice. A study published in The Journal of Diabetes examined this association and evaluated whether plasma volume mediates the observed risk. After expansion once, the abbreviation CHF-T2DM is used consistently.

The analysis combined two electronic medical record datasets: the Medical Information Mart for Intensive Care (MIMIC) database and the Tianjin Heart Failure with Integrated Treatment (TJHFIT) database. A total of 7,997 adults with CHF-T2DM were included. The MIMIC dataset contributed 6,112 adults, of whom 2,241 received insulin therapy and 3,871 did not. The TJHFIT dataset contributed 1,885 adults, with 911 receiving insulin therapy and 974 not receiving insulin therapy. Propensity score matching at a 1:2 ratio balanced demographics, comorbidities, and diabetes severity across groups.

Conditional logistic regression showed that insulin therapy was associated with higher in-hospital mortality. Odds ratios reached 1.37 (95% confidence interval 1.14–1.63) in MIMIC and 2.56 (95% confidence interval 1.53–4.27) in TJHFIT. Insulin therapy was also linked to a greater likelihood of a positive change in ePVS, with odds ratios of 1.11 in MIMIC and 1.37 in TJHFIT.

Restricted cubic spline modeling demonstrated nonlinear associations between ePVS and in-hospital mortality. Mediation analysis indicated that ePVS at discharge accounted for 12.0% of the insulin-associated mortality risk in MIMIC and 13.2% in TJHFIT.

These results suggest that plasma volume expansion contributes to the higher in-hospital mortality observed with insulin therapy in adults with CHF-T2DM.