The study is a retrospective cohort. It has been approved by our university’s institutional review board, which has a subcontract with our hospital. No patient consent was required by the IRB due to the de-identified nature of the data.
This study was conducted in an urban setting at a Level I trauma center. The county population of approximately 800,000 residents is made up of 68.4% Caucasian, 13.9% Asian, 13.6% Hispanic, and 9.1% African-American residents. The county occupies 323 square miles with a combination of cities and suburban communities. The emergency medical services (EMS) system is 2-tiered, comprised of a combination of paid and volunteer basic life support (BLS) units and paid hospital-based Advanced Life Support (ALS) that contain 2 paramedics per unit. Based out of our hospital, there are 8 BLS units and 6 ALS units that respond to approximately 30,000 dispatches per year, 6,500 of these being treated by ALS.
In our system, glucose is administered in the form of 50% dextrose containing 25 grams for patients over the age of21. In patients over the age of 2, 25% dextrose is given. One hundred mg of thiamine can be administered either intramuscularly or intravenously. We have glucometers on all ALS units with protocols for dextrose replacement if the glucometers indicates capillary blood glucose (CBG) is less than 70mg/dl.
As no previous studies have shown the incidence of Wernicke’s induction with glucose administration, we selected a consecutive sample of 242 patients who were treated by ALS for hypoglycemia between May 1, 2008 and August 11, 2009.
The search terms “hypoglycemia” and “IV glucose” were used in our electronic medical record (EMR) (www.emscharts.com
, Atlanta, Georgia). This database was cross-referenced with the ED database, Sunrise Clinical Manager (SCM) (Eclipsys Corporations, Atlanta, Georgia) to investigate outcomes data.
All patients who were administered thiamine received 100 mg IV. Patients received glucose in the form of D50 if above 21 and D25 if above 2 years of age. No patients in our study were less than 2.
In choosing outcome measures we reviewed case reports of Wernicke’s encephalopathy for signs and symptoms that might be captured in routine EMS charting. Our review identified abnormal vital signs being induced by the carbohydrate load of glucose as an early predictor of Wernicke’s encephalopathy. These endpoints are routinely measured by prehospital care providers. Opthalomoplegia and ataxia are not documented routinely nor are they taught in our paramedic programs, but the Glasgow Coma Scale (GCS) has been shown to have high inter-rater reliability, is predictive of changes in mental status, and is routinely captured in EMS charts. GSC was our primary outcome with changes in heart rate, respiratory rate, and blood pressure as secondary outcomes. We also evaluated rates of discharge and reentry into the 911 system for patients who refused transport to the hospital. We estimated that a decrease in GCS of > 1, an increase in systolic blood pressure (SBP) > 10mmHG, and a heart rate increase by 20 to indicate a clinically significant change.
We calculated simple means and standard deviations of health indicators were calculated at each individual time point for patients administered and not administered thiamine. Means and standard deviations of percent changes in these health indicators were also calculated.
We used linear models to examine the effects of administration of thiamine (versus not) on changes in health indicators, adjusting for age and gender. Age- and gender-adjusted mean percent changes are presented along with 95% confidence intervals. We conducted sensitivity analyses excluding individuals who were administered Thiamine after D50 from the analyses described.
Because percent change in GCS was not normally distributed, we used a non-parametric analysis of variance as a sensitivity analysis. Specifically, we used a Kruskal-Wallis test to examine differences between the Thiamine and non-Thiamine groups in percent change in GCS. The significance of the effect of Thiamine derived from the Kruskal-Wallis test was similar to that reported from the linear model (data not shown).
We used the SAS software (SAS system for Windows, version 9.1.3; SAS Institute Inc, Cary, North Carolina) for all analyses.