Objectives: No consensus standard is established for the workup of diabetic patients with moderate hyperglycemia [fingerstick glucose of 11.00-33.00 mmol/l (200-600 mg/dl)]. We measured the incidence of serious electrolyte abnormalities in patients with moderate hyperglycemia and attempted to identify a subset of these patients who can safely forego routine electrolyte testing.
Methods: Prospective cohort study in two affiliated emergency departments.
Inclusion criteria: Moderate hyperglycemia.
Exclusion criteria: new onset diabetes or renal failure. The composite outcome was defined as an abnormal potassium K >5.00 or <3.50 mmol/l or diabetic ketoacidosis. Univariate analyses were performed using t-test and Fisher's exact test (alpha=0.05, two-tails). Logistic regression models were generated to identify variables that could predict the composite outcome.
Results: Three hundred and ninety-nine adults were enrolled (294 type II, 38% men). Mean age was 59+/-15 (20-97 years) years. The incidence of the composite outcome was 22% (95% confidence interval, 17.8-26.0%, n=86). The univariate analysis identified potassium-altering medications and insulin use as risk factors for the composite outcome. Logistic regression analysis identified potassium-altering medications as an increased risk (relative risk: 1.64, 95% confidence interval, 1.14-2.37) and use of oral hypoglycemics as a decreased risk (relative risk: 0.62, 95% confidence interval, 0.43-0.90) of the composite outcome. Cross-validation of the model demonstrated poor sensitivity (76%) and even worse accuracy (51%).
Conclusion: We failed to identify any subgroup of patients with moderate hyperglycemia who can be safely excluded from routine electrolyte testing. We recommend routine electrolyte testing for all moderate hyperglycemia patients in the emergency department.