GH Murata, Albuquerque VAMC; KD Adam, Albuquerque VAMC; RM Hoffman, Albuquerque VAMC; CS Wendel, Tucson VAMC; C Dalton, Tucson VAMC; MT Montagnini, Phoenix VAMC; SU Bokhari, Phoenix VAMC; JH Shah, Tucson VAMC; WC Duckworth, Phoenix VAMC

Objectives: Maintaining effective glycemic control with insulin requires frequent blood glucose monitoring for hypo- and hyperglycemic events. The purpose of this study was to describe the frequency of and risks for hypo- and hyperglycemic events in insulin-treated, type II diabetic veterans.

Methods: Subjects were randomly selected from pharmacy records at the Albuquerque, Tucson and Phoenix VAMCs. Patients were eligible for this study if, in the preceding 2 months, they received no new prescriptions for oral hypoglycemic agents and insulin doses were not increased by more than 10 units or 15%. They were instructed to monitor their blood glucose before breakfast, lunch and dinner and at bedtime for 8 consecutive weeks. Hemoglobin A1c (A1c) was measured at 4 and 8 weeks. Mean blood glucose at each testing time was considered a measure of glycemic control and the coefficient of variation (CV) was considered a measure of glucose lability. Logistic regression was used to identify patients at risk for hypo- (BG <=60 mg/dL) and hyperglycemic (BG>=400 mg/dL) events.

Results: Seventy-four subjects successfully completed the protocol. The mean self-measured BG was 172+36 mg/dL. Hypoglycemia occurred at least once in 63.5% of subjects and was detected in 2.03% of samples. Hyperglycemia occurred in 37.8% of subjects and was detected in 0.93% of samples. Repeated-measures ANOVA showed that mean BG steadily increased throughout the day (P<0.001) and was highest at bedtime (198+46 mg/dL). The glucose CVs varied by time of day (P<0.001) and were highest before lunch (36.0+9.9%). We compared subjects who had at least one pre-lunch reading <=60 mg/dL with those without pre-lunch hypoglycemia. Hypoglycemic patients had a significantly lower pre-lunch mean BG (155+34 vs 177+47 mg/dL; P=0.02) and higher pre-lunch glucose CV (40.8+9.3% vs 32.1+8.7%; P<0.001), but no differences were found in the 4- or 8- week A1c levels. The only risk factor for pre-lunch hypoglycemia was a high glucose CV. Subjects who had at least one bedtime BG >=400 mg/dL were also compared to those without bedtime hyperglycemia. Hyperglycemic subjects had a significantly higher bedtime mean BG (248+45 vs 179+30 mg/dL; P<0.001), a higher bedtime glucose CV (37.1+9.7% vs 31.2+7.0%; P=0.01), a higher 4-week A1c (8.8+1.5% vs 7.6+1.1%; P<0.001) and a higher 8-week A1c (8.5+1.5% vs 7.3+1.1%; P<0.001). A high bedtime mean glucose and glucose CV were independent risk factors for hyperglycemia.

Conclusions: The frequency of hypoglycemia was high in insulin-treated, type II diabetic veterans. Hypoglycemia occurred most often before lunch and was most closely correlated with a high glucose CV. On the other hand, hyperglycemia occurred most frequently at bedtime, reflected both higher mean values and a high glucose CV, and was correlated with poor long-term control.

Impact: Future studies should determine whether hypo- and hyperglycemic events are related to biologic variation or a mismatch between caloric intake and insulin dosing. If there is a mismatch, then a substantial number of veterans may need additional dietary counseling.