Impact of acute hyperglycemic crisis episode on survival in individuals with diabetic foot ulcer using a machine learning approach

Abstract

Objective: The outcome of DFUs concomitant with HCE remains unknown. This study aimed to investigate mortality rates and identify risk factors of mortality in patients with DFUs-HCE.

Methods: 27 inpatients with DFUs-HCE were retrospectively enrolled in a cohort design, they were compared to 93 inpatients with DFUs in a city designated emergency center, between January 2016 and January 2021. After a 6-year followed-up, clinical characteristic, amputation and survival rates were compared. Extreme gradient boosting was further used to explore the relative importance of HCE and other risk factors to all-cause mortality in DFUs.

Results: Patients with DFUs-HCE were more likely to havedementia, acute kidney injury and septic shock, whereas DFUs were more likely to have diabetic peripheral neuropathy and ulcer recurrence (P<0.05). No significant difference was observed on the amputation rate and diabetes duration. Both Kaplan-Meier curves and adjusted Cox proportional model revealed that DFUs-HCE was associated with a higher mortality compared with DFUs (P<0.05). HCE significantly increased the risk of mortality in patients with DFUs (hazard ratio, 1.941; 95% CI 1.018-3.700; P = 0.044) and was independent from other confounding factors (age, sex, diabetes duration, Wagner grades and Charlson Comorbidity Index). The XGBoost model also revealed that HCE was one of the most important risk factors associated with all-cause mortality in patients with DFUs.

Conclusions: DFUs-HCE had significantly lower immediate survival rates (first 1-6 month) than DFUs alone. HCE is an important risk factor for death in DFUs patients.

Keywords: diabetic foot ulcer; hyperglycemic crisis episode; machine learning; mortality rates; risk factors of mortality.