Perioperative Hypoglycemia in Diabetic Patients Undergoing Vitrectomy: a Masked CGM Prospective Observational Study

Patients with diabetes are at increased risk of perioperative glycemic disturbances due to preoperative fasting, the surgical stress response, and the use of insulin or oral hypoglycemic agents. Hypoglycemia, in particular, is an important clinical concern associated with neurological injury and cardiovascular complications. In routine practice, intraoperative glucose monitoring often relies on intermittent fingerstick testing or blood gas analysis, and these intermittent methods may fail to detect hypoglycemic episodes that occur during and immediately after surgery. Continuous glucose monitoring (CGM), which estimates glucose concentrations from interstitial fluid, allows continuous tracking of glycemic trends and may provide a more precise assessment of glycemic variability and hypoglycemia in the perioperative period. Although current guidelines recommend periodic intraoperative glucose measurement in patients with diabetes-especially those receiving insulin-glucose monitoring is frequently omitted during relatively short and stable ophthalmic procedures such as vitrectomy. Because vitrectomy is commonly performed in patients with diabetes, perioperative hypoglycemia in this population may go undetected. This is a single-center, prospective, observational cohort study conducted in patients with diabetes undergoing pars plana vitrectomy (TPPV). A masked CGM device (FreeStyle Libre 2®, an approved continuous glucose monitor) is applied for observational purposes only and does not influence clinical care. The sensor is placed on the upper arm before surgery. After an approximately one-hour warm-up period, glucose data are collected at 15-minute intervals from the completion of CGM warm-up until discharge from the post-anesthesia care unit (PACU); this interval defines the perioperative period for data collection. Throughout this period the CGM operates in a masked mode-no reader is provided, the device is not linked to the participant's smartphone, and alarms are