Effect of early versus delayed or no insulin therapy for moderate hyperglycemia in ICU patients: A target trial emulation.

Abstract

Hyperglycemia is common in intensive care unit (ICU) patients and associated with increased mortality. Despite several randomized clinical trials within glucose management in critical illness, the effects of insulin treatment, and its timing, in patients with moderate hyperglycemia remain uncertain. In acutely admitted ICU patients with moderate hyperglycemia, does insulin treatment improve clinical outcomes and does timing of insulin initiation matter? We performed a multicenter target trial emulation using targeted minimum loss-based estimation. We included patients from four ICUs at Karolinska University Hospitals, Solna, Huddinge, from 2010 to 2021. We assessed the effect of early insulin treatment compared with delayed or no insulin treatment on 30-day all-cause mortality in acutely admitted adult ICU patients with moderate hyperglycemia (10 - 13.9 mmol/L). We included 5,755 patients with moderate hyperglycemia; 3,149 (54.7%) received early insulin therapy (intervention group) and 2,606 (45.3%) received delayed or no insulin therapy (control group). The adjusted 30-day mortality was 22.1% (95% CI 21.1 - 23.1) for patients receiving insulin within 6 hours and 24.4% (95% CI 23.5 - 25.4) for patients with later or no insulin treatment, corresponding to an absolute risk reduction of 2.3% (95% CI 1.1 - 3.6). Absolute risk was reduced by 7.7%-points (CI 95% 6.5 - 8.9) when insulin was initiated within 1 hour of the first episode of moderate hyperglycemia and by 6.6%-points (CI 95% 5.4 - 7.8) with initiation within 2 hours. This effect was modified in several subgroups. In this target trial emulation, early insulin treatment of moderate hyperglycemia showed reduced mortality in acutely admitted adult ICU patients - particularly when insulin was initiated within 2 hours following hyperglycemia. These findings support further investigation in timing of insulin initiation.