Validation Study of a Web-Based Periprocedural Diabetes Management Algorithm (DIAPI) in an Endoscopy Unit

Periprocedural diabetes management is challenging in endoscopy units due to modified diets, fasting, and bowel preparations. DIAPI (Diabetes Management in Peri-Intervention), a web-based tool, generates personalized treatment directives based on patient and procedure variables. This study evaluated DIAPI's reproducibility and noninferiority compared with physician orders.

This open-label interventional trial with sequential group assignment included adults with diabetes who were undergoing nonurgent procedures. In phase 1, DIAPI's reproducibility was assessed by comparing directives generated from clinical data collected by different health care professionals (physicians compared with nurses). Patients in the phase 1 group managed their diabetes according to physicians' orders (control arm). Patients in the phase 2 group followed DIAPI-generated directives based on clinical data collected by endoscopy unit nurses (intervention arm). Noninferiority (15% margin; α = 0.05) was tested for in-target glucose levels (70-180 mg/dL) during the endoscopy unit stay.

Baseline characteristics were similar between patients in phase 1 (n = 75 patients) and phase 2 (n = 98 patients): median age 65 versus 66 years old and 49.3% versus 51.0% treated with insulin. Most patients had type 2 diabetes (85%) and underwent colonoscopy (59%). DIAPI directives were reproducible in 83% and were all endorsed by endocrinologists. Efficacy was noninferior in the DIAPI arm, with 79.4% of patients achieving in-target glucose levels compared with 70.0% in the physician arm (+9.4%; 95% CI -0.018 to 1.0; P = 0.0002), thus favoring DIAPI. No adverse effects beyond mild dysglycemias were observed in either arm.

The DIAPI tool offers an effective, safe, reliable, and potentially cost-efficient solution for simplifying periprocedural diabetes management in endoscopy units.