Diabetes in urban African-Americans. VI. Utility of fasting or random glucose in identifying poor glycemic control.

ABSTRACT

OBJECTIVE: African-Americans have an increased prevalence of both diabetes and diabetes complications, creating an imperative for improved metabolic control. Because American Diabetes Association guidelines recommend that action be taken when HbA1c is > 8.0%, but access to rapid-turnaround HbA1c assays remains limited, we tested the utility of fasting and random plasma glucose cutoffs as indicators of HbA1c > 8.0%. RESEARCH DESIGN AND METHODS: Using receiver operating characteristics (ROC) analysis, we evaluated the sensitivity, specificity, and predictive value of fasting and random plasma glucose measurements in identifying an HbA1c > 8.0% (fasting n = 974, random n = 552). The population studied was predominantly African-American, middle-aged, and non-insulin-dependent. RESULTS: Fasting plasma glucose was a significant indicator of HbA1c > 8.0%, both in the whole group and in subgroups for diet, sulfonylureas, and insulin; the corresponding areas under the ROC curve were 0.87, 0.90, 0.87, and 0.84, respectively (all P < 0.0001). A fasting plasma glucose cutoff of > 9.2 mmol/l (165 mg/dl) provided a sensitivity of 80% and a specificity of 83% for the whole group and a 77% positive predictive value. Random plasma glucose was also a good indicator of HbA1c > 8.0%, both in the whole group and in subgroups for diet, sulfonylureas, and insulin; the corresponding areas under the ROC curve were 0.85, 0.91, 0.85, and 0.77, respectively (all P < 0.0001). A cutoff > 9.8 mmol/l (177 mg/dl) provided a sensitivity of 78% and a specificity of 77% for the whole group and a 78% positive predictive value. Overall, a plasma glucose > 11.1 mmol/l (200 mg/dl) identified an HbA1c > 8.0% with a predictive value of approximately 90% if done while fasting and a predictive value of approximately 80-85% if random. The utility of both fasting and random plasma glucose cutoffs was subsequently confirmed in a prospective study of another 2,309 and 1,396 patients, respectively. CONCLUSIONS: Although glucose levels cannot replace HbA1c determinations, measurement of fasting or random plasma glucose may be used during a clinic visit to identify poorly controlled type 2 patients with reasonable certainty and allow timely patient education and therapeutic intervention.