Background: This study aimed to assess the accuracy and utility of high-resolution continuous glucose recording in patients with traumatic brain injury (TBI) and to establish whether a relationship exists between the cumulative amplitude and duration of hyperglycemia and outcome after TBI.
Methods: Glucose data for 56 TBI patients were collected continuously at 5-min intervals. The degree and duration of hyperglycemia above treatment thresholds were calculated as "glucose times time dose" (GTD; mg/dL d) using continuous recordings (GTD) for early stage (first 3 days). Long-term neurological functional outcome was assessed using the extended Glasgow Outcome Scale (GOSE). Receiver operating characteristic (ROC) curves were constructed to determine the predictive values of GTD, percentage readings, mean, and range of glucose for in-hospital mortality and GOSE.
Results: All measurements of GTD were statistically significantly higher in the group that died. GTD of glucose >150 and glucose >180 had a high-predictive power for in-hospital mortality (areas under the ROC curve [AUC] = 0.917; 95 % CI, 0.837-0.998 and 0.876; 95 % CI, 0.784-0.967, respectively) and demonstrated significantly higher predictive power for mortality when compared with %reading >150 and %reading >180, respectively (p < 0.05). GTD of glucose >150 also had a significantly higher predictive power for mortality than mean glucose and range of glucose. GTD of glucose >150 and glucose >180 also had a high-predictive power for poor outcome (areas under the ROC curve [AUC] = 0.913; 95 % CI, 0.843-0.983 and 0.858; 95 % CI, 0.760-0.956, respectively).
Conclusions: Continuous collection of glucose recordings is more reliable and accurate than routine discontinuous recordings. Assessing both the duration and the amplitude of the episodes using continuous collection of glucose data helps in better predicting outcomes than the total duration of episodes.