Prediction of Hypoglycemia During Admission of Non-Critically Ill Patients: Results from the MENU Study

Abstract

AbstractThe objective of the work was to study admission parameters associated with an increased incidence of hypoglycemia during hospitalization of non-critically ill patients. Included in this cross-sectional study were patients admitted to internal medicine units. The Nutritional Risk Screening 2002 (NRS2002) was used for nutritional screening. Data recorded included admission serum albumin (ASA) and all glucose measurements obtained by the institutional blood glucose monitoring system. Neither of these are included in the NRS2002 metrics. Hypoalbuminemia was defined as ASA<3.5 g/dl. Patients were categorized as hypoglycemic if they had at least one documented glucose≤70 mg/dl during the hospitalization period. Included were 1342 patients [median age 75 years (IQR 61–84), 51.3% male, 52.5% with diabetes mellitus, (DM)], who were screened during three distinct periods of time from 2011–2018. The incidence of hypoglycemia was 10.8% with higher rates among DM patients (14.6 vs. 6.6%, p<0.001). Hypoglycemia incidence was negatively associated with ASA regardless of DM status. Multivariable regression showed that ASA (OR 0.550 per g/dl, 95% CI 0.387–0.781, p=0.001) and positive NRS2002 (OR 1.625, 95% CI 1.072–2.465, p=0.022) were significantly associated with hypoglycemia. The addition of hypoalbuminemia status to the NRS2002 tool improved the overall sensitivity from 0.55 to 0.71, but reduced specificity from 0.63 to 0.46. The negative predictive value was 0.93. Our data suggest that the combination of positive malnutrition screen and hypoalbuminemia upon admission are independently associated with the incidence of hypoglycemia among non-critically ill patients, regardless of diabetes mellitus status.