Relationship between hyperglycemia and outcome in children with severe traumatic brain injury

Abstract

Objective: To determine the relationship between hyperglycemia and outcome in infants and children after severe traumatic brain injury.

Design: Retrospective review of a prospectively collected Pediatric Neurotrauma Registry.

Setting and patients: Children admitted after severe traumatic brain injury (postresuscitation Glasgow Coma Scale ≤ 8) were studied (1999-2004). A subset of children (n = 28) were concurrently enrolled in a randomized, controlled clinical trial of early hypothermia for neuroprotection.

Interventions: Demographic data, serum glucose concentrations, and outcome assessments were collected.

Methods and main results: Children (n = 57) were treated with a standard traumatic brain injury protocol. Exogenous glucose was withheld for 48 hrs after injury unless hypoglycemia was observed (blood glucose <70 mg/dL). Early (first 48 hrs) and Late (49-168 hrs) time periods were defined and mean blood glucose concentrations were calculated. Additionally, children were categorized based on peak blood glucose concentrations during each time period (normal, blood glucose <150 mg/dL; mild hyperglycemia, blood glucose ≤ 200 mg/dL; severe hyperglycemia, blood glucose >200 mg/dL). In the Late period, an association between elevated mean serum glucose concentration and outcome was observed (133.5 ± 5.6 mg/dL in the unfavorable group vs. 115.4 ± 4.1 mg/dL in favorable group, p = .02). This association continued to be significant after correcting for injury severity, age, and exposure to insulin (p = .03). Similarly, in the Late period, children within the severe hyperglycemia group had decreased incidence of good outcome compared to children within the other glycemic groups (% good outcome: normal, 61.9%; mild hyperglycemia, 73.7%; severe hyperglycemia, 33.3%; p = .05). However, when adjusted for exposure to insulin, this relationship was no longer statistically significant.

Conclusions: In children with severe traumatic brain injury, hyperglycemia beyond the initial 48 hrs is associated with poor outcome. This relationship was observed in both our analysis of mean blood glucose concentrations as well as among the patients with episodic severe hyperglycemia. This observation suggests a relationship between hyperglycemia and outcome from traumatic brain injury. However, only a prospective study can answer the important question of whether manipulating serum glucose concentration can improve outcome after traumatic brain injury in children.

Figure 1

Flow diagram describing the relationship between glycemic stratification in the Early and Late time periods [definitions of glycemic strata: normal (NG) – peak glucose < 150 mg/dL; mild hyperglycemia (MHG) – peak glucose ≤ 200 mg/dL; severe hyperglycemia (SHG) – peak glucose > 200 mg/dL].

Figure 2

Mean serum glucose concentrations in the Early period were similar in children with favorable (light grey bar) and unfavorable outcomes (dark grey bar)(133.3 ± 5.2 vs. 135.7 ± 7.3, respectively, p = 0.72) at 6 months. In the Late period, an association between elevated mean serum glucose concentrations and poor outcome was observed (133.5 ± 5.6 vs. 115.4 ± 4.1, p = 0.02, Figure 2). Neither relationship was altered after correction for multiple covariates.

Figure 3

No association between outcome and peak serum concentrations in the Early period (dark grey bars) was observed (% favorable outcome: NG – 60%, MHG – 54.5%, SHG – 60%, p = 0.80). In the Late period (light grey bars), children within the SHG had decreased incidence of good outcome compared to children within the other glycemic groups (% good outcome: NG – 61.9%, MHG – 73.7%, SHG – 33.3%; p = 0.05, *denotes statistical significance between the NG and SHG groups [p = 0.05]). After controlling for insulin use, the effects of MHG and SHG on outcome were relatively strong but not statistically significant compared to NG (p = 0.323 and 0.272, respectively).