Relationship between blood glucose levels and length of hospital stay in patients with acute pancreatitis: An analysis of MIMIC‐III database

Abstract

Abstract We aimed to investigate the effect of blood glucose levels on length of stay (LOS) in patients hospitalized with acute pancreatitis (AP). We retrospectively collected clinical data of patients diagnosed with AP from the Medical Information Mart for Intensive Care III (MIMIC‐III) database. Dose–response analysis curves of restricted cubic spline (RCS) function and multivariate logistic regression models were used to confirm the relationship between blood glucose levels and LOS. A total of 3656 patients with AP were included according to the inclusion and exclusion criteria. According to RCS, all patients were divided into three groups, namely the less than 68 mg/dl group, the 68‐104 mg/dl group, and the >104 mg/dl group. RCS showed a significant nonlinear correlation between blood glucose levels and LOS (p < 0.001). Multivariate logistic regression revealed a 53% higher risk of LOS greater than or equal to 2 days (adjusted odds ratio [aOR] = 1.53, 95% confidence interval [CI] 1.24–1.89, p < 0.001), a 114% higher risk of LOS greater than or equal to 5 days (aOR = 2.14, 95% CI 1.86–2.47, p < 0.001), and a 130% higher risk of LOS greater than or equal to 7 days (aOR = 2.30, 95% CI 1.97–2.69, p < 0.001) in patients with glucose levels greater than 104 mg/dl than patients with glucose levels 68–104 mg/dl. The risk of LOS greater than or equal to 7 days was higher in patients with blood glucose less than 68 mg/dl than patients with glucose levels 68–104 mg/dl (aOR = 1.45, 95% CI 1.02–2.05, p = 0.040). In addition, we observed similar results in many subgroups. Our findings suggest that either hyperglycemia or hypoglycemia increase LOS in patients hospitalized with AP. For hospitalized patients with AP, blood glucose control in a reasonable range of 68–104 mg/dl is required.