Blood Glucose Range for Hyperglycemic PICU Children With Primary Neurologic Diagnoses: Analysis of the Heart and Lung Failure-Pediatric Insulin Titration (HALF-PINT) Trial.

Objectives: To compare two blood glucose (BG) ranges in critically ill children with and without primary neurologic diagnoses in the Heart and Lung Failure-Pediatric Insulin Titration trial (HALF-PINT; ClinicalTrials.gov Identifier NCT01565941).

Design: Non-prespecified post hoc analysis.

Setting: Thirty-one PICUs in the United States, and one in Canada.

Patients: Non-diabetic children enrolled from April 2012 to September 2016 with cardiovascular or respiratory failure and hyperglycemia. Patients in the neurologic subgroup had primary neurologic diagnoses on ICU admission.

Interventions: Patients were randomized to insulin infusion to target lower-BG (80-110 mg/dL; 4.4-6.1 mmol/L) or higher-BG (150-180 mg/dL; 8.3-10 mmol/L).

Measurements and main results: Primary diagnosis (neurologic vs. non-neurologic), daily BG and insulin values, outcomes (number of PICU-free days through day 28 and 1-y post-PICU discharge adaptive behavior composite score of Vineland Adaptive Behavior Scales, Second Edition). Of 698 patients analyzed, 64 (30 lower-BG target, 34 higher-BG target) had primary neurologic diagnoses and 634 (319 lower-BG target, 315 higher-BG target) had non-neurologic diagnoses. Within the neurologic subgroup, patients in the lower-BG targeting group had fewer ICU-free days compared with those in the higher-BG targeting group (median 8.5 vs. 21.1 d), whereas there was no difference between BG groups in the non-neurologic subgroup (20.5 vs. 19.3 d; interaction p = 0.02). One-year adaptive behavior composite score was less favorable for the lower-BG targeting group in those with neurologic diagnoses (mean 63.3 vs. 87.6), but no different in those with non-neurologic diagnoses (81.9 vs. 78.4; interaction p = 0.02). Lower-BG targeting was associated with more hypoglycemia (< 60 mg/dL) in both diagnostic subgroups, with no differential effect across subgroups ( p = 0.47).

Conclusions: In this non-prespecified analysis of the HALF-PINT trial data, lower-BG targeting in hyperglycemic critically ill children with primary neurologic diagnoses was associated with unfavorable outcomes, while such BG targeting in those with non-neurologic diagnoses was not associated with adverse outcomes.