Background:
Critically ill infants and children often develop hyperglycaemia, which is associated with adverse outcome; however, whether lowering blood glucose concentrations to age-adjusted normal fasting values improves outcome is unknown. We investigated the effect of targeting age-adjusted normoglycaemia with insulin infusion in critically ill infants and children on outcome.

Methods:
In a prospective, randomised controlled study, we enrolled 700 critically ill patients, 317 infants (aged <1 year) and 383 children (aged ≥1 year), who were admitted to the paediatric intensive care unit (PICU) of the University Hospital of Leuven, Belgium. Patients were randomly assigned by blinded envelopes to target blood glucose concentrations of 2·8–4·4mmol/L in infants and 3·9–5·6mmol/L in children with insulin infusion throughout PICU stay (intensive group [n=349]), or to insulin infusion only to prevent blood glucose from exceeding 11·9mmol/L (conventional group [n=351]). Patients and laboratory staff were blinded to treatment allocation. Primary endpoints were duration of PICU stay and inflammation. Analysis was by intention to treat. This study is registered with ClinicalTrials.gov, number NCT00214916.

Findings:
Mean blood glucose concentrations were lower in the intensive group than in the conventional group (infants: 4·8 [SD 1·2]mmol/L vs 6·4 [1·2]mmol/L, p<0·0001; children: 5·3 [1·1]mmol/L vs 8·2 [3·3]mmol/L, p<0·0001). Hypoglycaemia (defined as blood glucose ≤2.2mmol/L) occurred in 87 (25%) patients in the intensive group (p<0·0001) versus five (1%) patients in the conventional group; hypoglycaemia defined as blood glucose less than 1·7mmol/L arose in 17 (5%) patients versus three (1%) (p=0·001). Duration of PICU stay was shortest in the intensively treated group (5·51 days [95% CI 4·65–6·37]vs 6.15 days [5·25–7·05], p=0·017). The inflammatory response was attenuated at day 5, as indicated by lower C-reactive protein in the intensive group compared with baseline (−9.75mg/L [95% CI −19·93 to 0·43]vs 8·97mg/L [−0·9 to 18·84], p=0·007). The number of patients with extended (>median) stay in PICU was 132 (38%) in the intensive group versus 165 (47%) in the conventional group (p=0·013). Nine (3%) patients died in the intensively treated group versus 20 (6%) in the conventional group (p=0·038).

Interpretation:
Targeting of blood glucose concentrations to age-adjusted normal fasting concentrations improved short-term outcome of patients in PICU. The effect on long-term survival, morbidity, and neurocognitive development needs to be investigated.

Funding:
Research Foundation (Belgium); Research Fund of the University of Leuven (Belgium) and the EU Information Society Technologies Integrated project “CLINICIP”; and Institute for Science and Technology (Belgium) (Fig 4).

Figure 4: Effect on inflammation and mortality (A) Mean changes from day 1 concentrations of C-reactive protein (CRP) in the conventional and intensive insulin groups for the first 5 days in PICU. Error bars indicate SE. p value was obtained by repeated measures ANOVA for overall significance of the difference in time course of CRP (time*treatment interaction). The p value for the change in CRP on day 5, calculated by Mann-Whitney U test, was 0·007. (B) Kaplan-Meier analysis depicting cumulative incidence of PICU death (%) for time (days) in PICU in the conventional and intensive insulin groups. The p value was obtained by log-rank testing. (Reprinted from Vlasselaers D, Milants I, Desmet L, et al. Intensive insulin therapy for patients in paediatric intensive care: a prospective, randomised controlled study. Lancet 373:547-556, 2009 with permission from Elsevier.)