Objective: To determine the optimal dose of insulin glargine needed to maintain glycemic control in patients undergoing conversion from intravenous regular insulin infusions to a subcutaneous insulin regimen.
Methods: Seventy-five hospitalized patients receiving continuous insulin infusions were randomized to receive 40%, 60%, or 80% of their total daily insulin requirement, calculated from the rate during the final 6 hours of the infusion, as insulin glargine at the time of conversion to a subcutaneous regimen. Prandial insulin aspart was added to the subcutaneous regimen when patients began oral intake, and the dosage was left to clinical judgment. Capillary blood glucose monitoring (CBGM) was performed before every meal and at bedtime. All CBGM values for the 24-hour period after conversion were collected.
Results: Three hundred ninety-two CBGM values were recorded and analyzed. The mean for all CBGM values during the 24-hour period after conversion to the subcutaneous insulin regimen was 151.9 +/- 42.5 mg/dL in the 40% group, 164.0 +/- 41.6 mg/dL in the 60% group, and 153.2 +/- 66.2 mg/dL in the 80% group (P = 0.66). The percentage of CBGM values in the predefined study target range (80 to 140 mg/dL) was 43.2%, 34.8%, and 48% in the 40%, 60%, and 80% groups, respectively (P = 0.09). Secondary analysis with use of a glycemic target of 80 to 150 mg/dL and removal of outliers resulted in CBGM values within that range in 58.7%, 44.4%, and 67.6% for the 40%, 60%, and 80% groups, respectively (overall, P = 0.001; 40% group versus the 60% group, P = 0.03; 60% group versus the 80% group, P = 0.0004; and 40% group versus the 80% group, P = 0.18).
Conclusion: Conversion from continuous insulin infusion to subcutaneously administered insulin glargine at a dose equal to 80% of the total daily insulin requirements resulted in the highest percentage of CBGM values in the glycemic target range of 80 to 150 mg/dL within the first 24 hours after regimen conversion in comparison with conversion at 40% and 60%, albeit the difference between the 40% and 80% groups was not statistically significant.