Impact of tight blood glucose control on atrial fibrillation in critically ill patients receiving early parenteral nutrition: an individual patient data meta-analysis of two large randomized controlled trials

Atrial fibrillation often occurs during critical illness. Tight glucose control with insulin (TGC) is known to reduce inflammation and oxidative stress and may alter atrial metabolism, which, together, could affect atrial fibrillation pathogenesis. Our group has previously shown that TGC reduced morbidity and mortality in a mixed medical/surgical critically ill patient population receiving early parenteral nutrition as part of the contemporary standard of care. We here hypothesized that TGC reduces atrial fibrillation in the intensive care unit (ICU). In this individual patient data meta-analysis of 2 randomized controlled trials (performed 2000–2001 and 2002–2005), we investigated the impact of TGC with insulin (targeting blood glucose 80–110 mg/dL) in comparison with tolerating hyperglycemia to 215 mg/dL (liberal glucose control [LGC]) on atrial fibrillation in mixed surgical (n = 1548) and medical (n = 1200) ICU patients admitted to a quaternary-care university hospital. Atrial fibrillation was further classified as new-onset or recurrent/persistent pre-existing atrial fibrillation. The primary endpoint was the impact of TGC on atrial fibrillation in ICU, determined via multivariable logistic regression analysis after adjusting for relevant baseline patient characteristics. Prespecified subgroup analyses were performed for patients with a history of diabetes mellitus, pre-existing atrial fibrillation, overall surgical admission, admission after cardiac surgery and for patients with an ICU-stay longer than 5 days, after assessing treatment heterogeneity via determination of interaction p-values. Atrial fibrillation in ICU occurred in 845/2639 patients (32.0%), 65.9% of which was new-onset atrial fibrillation. TGC had no impact on atrial fibrillation in ICU (adjusted OR 0.92 [0.77–1.11])(P = 0.40). TGC also did not affect new-onset atrial fibrillation (adjusted OR 0.92 [0.75–1.12])(P = 0.39). There was no treatment heterogeneity present for the pre-defined subgroups except for the subgroup of patients with history of diabetes mellitus (n = 390), in which atrial fibrillation was documented for 66/199 patients in the TGC-group (33.2%) and 80/191 patients in the LGC-group (41.9%), interaction P = 0.045). Atrial fibrillation in ICU was strongly associated with worse outcome. TGC in the context of early use of parenteral nutrition did not reduce atrial fibrillation during ICU-stay in this large mixed medical-surgical ICU cohort. A possible exception was noted for patients with a history of diabetes mellitus.