Evaluation of New Insulin Infusion Therapy with Potential Variables for Diabetes Mellitus ICU Patients

Insulin infusion therapy implementing a sliding scale is mainly used in Malaysia's ICU to control blood glucose (BG) levels. However, the sliding scale is a one-size-fits-all protocol, which may cause hypoglycemia events when trying to control hyperglycemia among ICU patients. Hence, this study evaluates two newly developed protocols with potential variables of insulin sensitivity (SI), a fraction of inspired oxygen (FiO
2
), and mean arterial pressure (MAP) to improve current insulin infusion therapy with personalized protocols and reduce the hypoglycemia event. Eighty-one diabetes mellitus patients who stayed in the ICU and received insulin infusion therapy were selected for this study. Two new protocols were developed, and the BG target was to be controlled within 8.0-10.0 mmol/L. Protocol A includes BG and SI, while Protocol B includes BG, SI, FiO
2
, and MAP. The outcomes from these protocols were compared with the performance of the clinical protocol. Protocol A resulted in 0.01 % of severe hypoglycemia events (BG < 2.2 mmol/L). Additionally, protocol A reduced severe hypoglycemia by 0.03% compared to the current protocol. Meanwhile, Protocol B had a similar percentage measurement of severe hypoglycemia events to the clinical protocol (0.04 %). Both protocols were able to control the blood glucose level within the target range with higher percentages of 28.45% and 28.22%, respectively, compared to the current clinical protocol (25.30%). Nevertheless, Protocol A and Protocol B had hyperglycemia events (BG > 10.0 mmol/L) slightly higher than clinical protocol, with percentage differences of 1.13% and 2.85%, respectively. This is because the insulin rate applied in both protocols was lower than the clinical protocol. The median insulin rate for Protocol A and Protocol B was the same, which was 1.2 U/hr, while the clinical protocol was 1.5 U/hr. The BG outcomes including median BG per-patient, percentage of severe hypoglycemia, hypoglycemia, hyperglycemia, and the BG within the target range, from Protocol A and Protocol B were not significantly different from the clinical protocol, with a p-value > 0.05. The potential variables introduced in the new protocols are able to reduce severe hypoglycemia events. However, they could not reduce the hyperglycemia events as both protocols provided lower insulin doses than the clinical protocol. Developing a personalized protocol is crucial, and more trials with large numbers of subjects are needed to properly design tight glycemic control for managing BG levels.