Background Studies of blood glucose (BG) control during Paediatric Intensive Care Unit (PICU) admission have demonstrated worse outcomes with hyper- and hypoglycemia. The majority of these studies have used intermittent blood sampling protocols, which are inherently prone to sampling frequency bias . Admission BG is a single value, and is therefore immune to sampling frequency bias. The relationship between clinical outcomes and admission BG has been described in critically ill adults , but this relationship has not been previously investigated in critically ill children.
Aims In a cohort of PICU patients to describe:
The relationship between admission BG and PICU mortality.
The prevalence of admission hypoglycaemia.
Setting Mixed medical and surgical PICU in a tertiary children’s hospital.
Methods Retrospective, analytical, cohort study. BG values obtained from blood gas electronic database. Admission glucose defined as first BG within one hour of admission. Readmissions were excluded. Mortality rate was calculated for each decile of admission BG. Proportion of hypoglycaemia was calculated for whole cohort, and by age and surgical status. The definitions of hypoglycaemia: severe <=2.2 mmol/L; moderate <4.0 mmol/L.
Results 3129 consecutive patients were included. PICU mortality by decile of BG demonstrated a U-shaped curve: higher mortality risk at low and high values (Figure 1).
Overall prevalence of admission hypoglycemia: Moderate: 7.6%. Severe: 1.3% Prevalence of admission hypoglycemia was inversely related to age, with highest prevalence in neonates. Admission hypoglycemia was more common in non-surgical patients, compared with surgical patients in all age groups. Highest prevalence of hypoglycaemia was in non-surgical neonates: 23.5%, 2.9% – moderate and severe hypoglycemia, respectively.
Conclusions Previous studies have focused on the clinical associations of hyperglycaemia. This study shows that admission hypoglycemia is equally important as hyperglycaemia, in terms of PICU mortality. The higher prevalence of hypoglycaemia in non-surgical neonatal patients may reflect reduced or exhausted glycogen stores. Clinicians should be aware of this marker of raised mortality risk when admitting patients to PICU.
Plunkett et al. Observational Studies of Glucose Homeostasis are Susceptible to Sampling Frequency Bias. Pediatric Critical Care Medicine 2013
Evans et al. Assessing the relationship between admission glucose levels, subsequent length of hospital stay, readmission and mortality. Clinical Medicine 2012