To compare the efficacy of an extended insulin regimen using correction of hyperketonemia as endpoint with a more conventional regimen in the treatment of diabetic ketoacidosis.


A total of 22 patients admitted to a Belfast teaching hospital with clinical and biochemical features of diabetic ketoacidosis (pH < 7.25 and/or bicarbonate < 16 mmol/l) were randomized to either conventional or extended insulin regimens. In the conventional regimen, insulin was administered at 5 U/h until near-normoglycemia (blood glucose ≤ 10 mmol/l) and then administered at a reduced rate until clinical recovery. In the extended regimen, administration of insulin at 5 U/h was continued beyond attainment of normoglycemia, until resolution of hyperketonemia (3-hydroxybutyrate < 0.5 mmol/l). Main outcome measures were 3-hydroxybutyrate and bicarbonate levels during the 24 h after attainment of near-normoglycemia.


After near-normoglycemia, correction of hyperketonemia was achieved earlier with the extended treatment (5.9 ± 0.8 vs. 21.8 ± 3.4 h, P = 0.0004 [mean ± SD]). The area under the curve of 3-hydroxybutyrate against time for 24 h after near-normoglycemia was reduced with the extended treatment (24.9 ± 3.8 vs. 55.9 ± 6.7 mmol·l−1·h−1P = 0.001). These differences remained statistically significant after adjustment for higher baseline levels of 3-hydroxybutyrate at near-normoglycemia in the extended treatment group. Bicarbonate levels at 6 and 12 h after near-normoglycemia were not significantly different between groups.


The extended insulin regimen, which was easy to implement at ward level, produced a more rapid resolution of ketosis than the conventional regimen.

This content is only available via PDF.