Objectives: The objectives were to evaluate the correlation between random glucose and hemoglobin A1c (HbA1c) in emergency department (ED) patients without known diabetes and to determine the ability of diabetes screening in the ED to predict outpatient diabetes.
Methods: This was a cross-sectional study at an urban academic ED. The authors enrolled consecutive adult patients without known diabetes during eight 24-hour periods. Point-of-care (POC) random capillary glucose and HbA1c levels were tested, as well as laboratory HbA1c in a subset of patients. Participants with HbA1c > or = 6.1% were scheduled for oral glucose tolerance test (OGTT).
Results: The 265 enrolled patients were 47% female and 80% white, with a median age of 42 years. Median glucose and HbA1c levels were 93 mg/dL (interquartile range [IQR] = 82-108) and 5.8% (IQR = 5.5-6.2), respectively. The correlation between POC and laboratory HbA1c was r = 0.96, with mean difference 0.33% (95% confidence interval [CI] = 0.27% to 0.39%). Glucose threshold > or = 120 mg/dL had 89% specificity and 26% sensitivity for predicting the 76 (29%) patients with abnormal HbA1c; > or = 140 mg/dL had 98% specificity and 14% sensitivity. The correlation between random glucose and HbA1c was moderate (r = 0.60) and was affected by age, gender, prandial status, corticosteroid use, and current injury. Only 38% of participants with abnormal HbA1c returned for OGTTs; 38% had diabetes, 34% had impaired fasting glucose/impaired glucose tolerance, and 28% had normal glucose tolerance.
Conclusions: ED patients have a high prevalence of undiagnosed diabetes. Although screening with POC random glucose and HbA1c is promising, improvement in follow-up with confirmatory testing and initiation of treatment is needed before opportunistic ED screening can be recommended.