Management of diabetes mellitus in hospitalized patients: Efficiency and effectiveness of sliding-scale insulin therapy

Study Objective. To determine the efficiency and effectiveness of current prescribing practices relative to short- and intermediate-acting insulins in the prevention or treatment of acute hyperglycemic episodes in hospitalized patients with diabetes mellitus or hyperglycemia, and to identify clinical findings that influence the effectiveness of insulin therapy in these patients. Design. Retrospective observational study Setting. University-affiliated hospital. Patients. Ninety consecutive adult inpatients who had orders placed for as-needed subcutaneous regular or lispro sliding-scale insulin. Measurements and Main Results. Medical records were reviewed for patients' clinical characteristics and responses to administered insulin that were recorded during each of the first 5 days of hospitalization in which sliding-scale insulin therapy was used. Despite the immediate or bedside availability of both computerized and manual means to record finger-stick blood glucose levels and insulin injections, uncertainties or missing information related to execution, timing, blood glucose levels, or insulin dose were present in approximately 30% of all anticipated points of care involving insulin. Ten episodes of hypoglycemia in six patients were associated with sliding-scale insulin. Appropriately timed, successive glucose measurements documented a decrement in elevated blood glucose values to within the target range of 90-130 mg/dl after 76 (12%) of 621 sliding-scale insulin injections. Glucose levels remained elevated, and insulin effects were therefore subtherapeutic after 523 injections (84%). Despite blood glucose levels that remained persistently elevated, corresponding adjustments in either the timing or the dose of insulin were made infrequently. Sliding-scale insulin regimens were never adjusted in 73 patients (81%). Through 5 days of therapy, the proportion of patients who attained good glycemic control ranged from 2-10% (mean 6%). The mode of overall glycemic control was poor, with 51-68% of patients in this category on any given day. Overall, treated diabetic and hyperglycemic patients were more likely to be poorly controlled than relatively well controlled. Conclusion. Our findings reveal outcomes associated with sliding-scale insulin that are widely variable, often ineffectual, and prone to deficiencies in monitoring, documentation, and prescribing soundness. Efforts to improve glycemic control in hospitalized patients are clearly needed.