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

1. Background and Rationale Patients with diabetes mellitus undergoing surgery are exposed to multiple factors that can destabilize glycemic control, including preoperative fasting, the neuroendocrine stress response to surgery and anesthesia, and the perioperative use of insulin or oral hypoglycemic agents. Among the resulting glycemic disturbances, hypoglycemia is of particular clinical concern because it can be associated with neurological injury and cardiovascular complications, and because its symptoms are frequently masked under general anesthesia and sedation. An unrecognized intraoperative or early postoperative hypoglycemic episode may therefore go undetected and untreated. In routine clinical practice, intraoperative glucose monitoring commonly relies on intermittent point-of-care testing (POCT), such as fingerstick capillary glucose measurement, or on blood gas analysis. These intermittent methods provide only discrete snapshots of glucose status and may fail to capture transient hypoglycemic episodes occurring between measurements. As a result, the true burden of perioperative hypoglycemia in surgical patients with diabetes is likely underestimated. Continuous glucose monitoring (CGM) measures interstitial fluid glucose concentrations and provides a continuous record of glycemic trends. CGM has emerged as a tool that may allow more precise assessment of glycemic variability and detection of hypoglycemia in the perioperative setting, and recent studies have reported its usefulness for this purpose. Current guidelines recommend periodic intraoperative glucose measurement (approximately every one to two hours) in patients with diabetes, particularly those receiving insulin. In practice, however, glucose monitoring is frequently omitted during relatively short and hemodynamically stable ophthalmic procedures such as pars plana vitrectomy (TPPV). HbA1c reflects chronic glycemic control but has limited ability to predict acute hypoglycemic events. Vitrectomy is notable for being performed in a high proportion of patients with diabetes, yet perioperative glycemic monitoring during such surgery remains limited. Consequently, hypoglycemia occurring intraoperatively or in the recovery room may not be adequately detected. There is a need to characterize the incidence and pattern of perioperative hypoglycemia in this population using a continuous monitoring approach. 2. Study Objectives The primary objective of this study is to estimate the incidence of perioperative hypoglycemia, defined as a CGM glucose value below 70 mg/dL (Level 1), in patients with diabetes undergoing vitrectomy, using masked CGM. Secondary objectives are to evaluate: The incidence of clinically significant hypoglycemia, defined as a CGM glucose value below 54 mg/dL (Level 2); Indices of glycemic variability during the perioperative period, including Time Below Range (TBR), maximum and minimum glucose values, and glucose excursion (maximum minus minimum glucose); Clinical risk factors associated with perioperative hypoglycemia, such as age, diabetes characteristics, HbA1c, insulin use, fasting duration, and operative time; The exploratory association between perioperative hypoglycemia and postoperative ophthalmic outcomes; and The exploratory, opportunistic concordance between standard-of-care POCT glucose values and time-matched CGM values. This study is intended to characterize the limitations of current perioperative glucose monitoring strategies and to provide foundational data for the design of future risk-based monitoring strategies and CGM-based interventional research. 3. Study Hypotheses The study is based on the following hypotheses: Perioperative hypoglycemia occurs in patients with diabetes undergoing vitrectomy. The occurrence of perioperative hypoglycemia is associated with clinical risk factors including age, diabetes characteristics, HbA1c, insulin use, fasting duration, and operative time. CGM can continuously record glycemic variability during the perioperative period, and the concordance between CGM values and opportunistically obtained standard-of-care POCT values can be assessed in an exploratory manner. 4. Study Design and Setting This is a single-center, prospective, observational cohort study. It is non-interventional with respect to clinical management: a masked CGM device is applied solely for observational data collection and does not influence clinical decision-making. No randomization or treatment allocation is performed. Glycemic management is conducted entirely according to the existing standard of care. 5. Study Device The study uses the FreeStyle Libre 2® continuous glucose monitor (Abbott Diabetes Care). This device is an approved continuous glucose monitoring system in Korea. In this study it is applied in a masked, observational manner rather than for therapeutic glucose management. The sensor records interstitial glucose at frequent intervals, and data are retrieved after the monitoring period for analysis. 6. Eligibility Criteria Inclusion criteria: Adults aged 19 years or older; Diagnosed diabetes mellitus; Scheduled to undergo vitrectomy; Able to provide written informed consent. Exclusion criteria: CGM sensor placement not feasible; Known allergy to the sensor; Inability to obtain valid CGM data; Inability to provide voluntary informed consent. 7. Enrollment and Informed Consent Recruitment and consent are performed by the principal investigator and study staff. During a preoperative outpatient visit, the study is explained in full, and participants are given adequate time to decide on participation before written informed consent is obtained. Participation intent is reconfirmed before study procedures begin. Each participant is assigned a screening number and a subject number, and identification is performed using the screening number, subject number, and initials. Participants are enrolled sequentially according to the time of consent, with no randomization. Each participant receives a copy of the information sheet and consent form, and is informed that consent may be withdrawn at any time without penalty. 8. CGM Application and the Perioperative Monitoring Period Before surgery, a CGM sensor is placed on the participant's upper arm. After sensor insertion, an approximately one-hour warm-up period elapses during which valid glucose data are not generated; this warm-up interval is excluded from analysis. Following completion of warm-up, CGM glucose data are collected at approximately 15-minute intervals from the end of warm-up until discharge from the post-anesthesia care unit (PACU). For the purposes of this study, the perioperative period for data collection is defined as the interval "from completion of CGM warm-up until PACU discharge." Within this interval, CGM glucose data, POCT glucose values, and surgical and anesthetic variables are collected. Data are obtained across the relevant care settings, including the operating room and the PACU. 9. Masking (Blinding) and Unblinding Procedures Single-blinding is applied. Blinding begins at the time of sensor placement and is maintained until PACU discharge and sensor removal (or until the end of the 14-day sensor wear period). During the primary data-collection interval (the perioperative period defined above), all clinical staff involved in care-anesthesiologists, the operating surgeon, and PACU nurses-have no access to CGM data. The CGM sensor is applied in a masked manner: no reader is distributed, the device is not linked to the participant's own smartphone, and alarms are disabled, so that real-time glucose values are not displayed. This is applied uniformly to all participants. Routine unblinding is performed collectively after data collection for all participants has been completed. Study staff log in to a dedicated research LibreView® account and download each participant's sensor data. The downloaded raw data are de-identified (matched by research number) and converted into an analysis dataset. This procedure is performed under the confirmation of the principal investigator, and the unblinding date, operator, and downloaded subject numbers are recorded in an unblinding log. Unblinded data are not transmitted to the clinical care team. Because this is a non-interventional observational study in which CGM does not affect clinical care and all glycemic management follows the standard of care (POCT), individual emergency unblinding is, in principle, not required. However, if a medical reason directly related to participant safety arises (for example, the need for post hoc review following a serious adverse event), the principal investigator may selectively unblind that participant's CGM data. In such cases, the reason, date, requestor, and approver of the emergency unblinding are recorded separately in the unblinding log and, where necessary, reported to the IRB. 10. Data Collected During the study, the following data are collected: Demographic variables; Underlying conditions, including diabetes type (Type 1 or Type 2), duration of diabetes, and insulin use; Surgical and anesthetic variables; CGM glucose data (masked, at approximately 15-minute intervals; clinicians do not view these data during the study period); POCT glucose values obtained according to standard care, including recovery-room (PACU) glucose measurements. At each time point, continuous CGM glucose data and intermittent standard-of-care glucose measurements are collected together. 11. Outcome Measures Primary outcome: Incidence of perioperative hypoglycemia (CGM glucose \< 70 mg/dL, Level 1). Definition: CGM glucose below 70 mg/dL. Determination: occurrence of at least one such episode during the perioperative period. Secondary outcomes: 1. Incidence of clinically significant hypoglycemia. Definition: CGM glucose below 54 mg/dL (Level 2). Determination: at least one episode during the perioperative period. 2. Opportunistic POCT-CGM concordance (exploratory). Standard-of-care POCT glucose values obtained at the discretion of clinical judgment are compared with CGM values at the same time point (within ±5 minutes) to assess concordance. Because the institution does not have a standardized intraoperative POCT protocol during TPPV, measurement timing is not fixed; this is therefore performed as an exploratory analysis rather than a formal outcome. 3. Glycemic variability indices: Time Below Range (TBR; \< 70 mg/dL and \< 54 mg/dL), maximum and minimum glucose values during the perioperative period, and glucose excursion (maximum minus minimum glucose). 4. Exploration of clinical risk factors for perioperative hypoglycemia: analysis of the association between hypoglycemia and factors such as age, diabetes characteristics, HbA1c, insulin use, fasting duration, and operative time. 5. Association between perioperative hypoglycemia and postoperative ophthalmic outcomes (exploratory): changes in best-corrected visual acuity (BCVA), and postoperative ophthalmic complication rates including recurrent vitreous hemorrhage, retinal redetachment, elevated intraocular pressure, and the need for additional surgery or procedures. Because the primary aim of the study is to estimate the incidence of perioperative hypoglycemia, ophthalmic outcomes are analyzed as exploratory endpoints. BCVA changes and ophthalmic complications are assessed from medical records of routine outpatient follow-up visits (for example, approximately one week and one to three months after surgery), without any additional study-specific visits. Hypoglycemia is classified according to ADA/Danne et al. criteria (Level 1: \< 70 mg/dL; Level 2: \< 54 mg/dL). 12. Statistical Analysis Continuous variables are summarized as mean ± standard deviation or as median with interquartile range, as appropriate, and categorical variables are presented as frequencies and percentages. The primary outcome, the incidence of perioperative hypoglycemia (CGM \< 70 mg/dL), is reported with a Wilson 95% confidence interval. Associated factors are evaluated using logistic regression analysis. Analyses are performed using SPSS, with a two-sided significance threshold of p \< 0.05. 13. Sample Size and Feasibility Given the institution's monthly surgical volume (approximately 50-60 vitrectomies per month, with an estimated proportion of patients with diabetes of approximately 40%), roughly 20 to 24 eligible participants per month can be identified. Accounting for consent and dropout rates over a 12-month study period, the target enrollment is set at 85 participants. The institution's surgical volume supports the feasibility of recruiting 85 participants within 12 months of IRB approval. The results are expected to inform the development of risk-based glucose monitoring strategies, more precise future sample-size calculations, and the design of CGM-based interventional studies. 14. Study Period and Follow-up The study period is one year (12 months) from the date of IRB approval. Postoperative ophthalmic outcomes are obtained from routine outpatient follow-up records without additional study visits. 15. Termination and Early Discontinuation If a participant withdraws consent, only data collected up to the time of withdrawal are included in the analysis, and no further data collection is performed. Participants may be discontinued if valid data cannot be collected or if follow-up becomes impossible. 16. Safety Monitoring This study involves a low level of risk. The only study-specific procedure beyond routine care is placement of the FreeStyle Libre 2® sensor on the upper arm to collect continuous glucose data; no additional invasive procedures are performed. The sensor is a widely used medical device, and local skin reactions such as mild redness or itching at the application site may occur but are generally mild and transient. There is no direct therapeutic benefit from study participation, but the study is expected to help characterize the limitations of hypoglycemia detection during ophthalmic surgery and recovery in patients with diabetes, and to provide foundational data for safer perioperative glucose monitoring strategies. The anticipated risk is very low and is outweighed by the potential benefit. An independent safety monitor serves as the protocol safety manager in accordance with the risk level of the study. The independent safety monitor receives reports on protocol adherence, deviations, and withdrawals from the investigators, reviews the completeness of data, and evaluates findings related to adverse events. Safety monitoring is conducted after enrollment of the first 15 cases and thereafter at six-month intervals, comparing case report forms against the protocol to review participant safety data. In the event of an unexpected serious medical occurrence during the study (death, a life-threatening event, hospitalization or prolongation of hospitalization, persistent or significant disability or functional impairment, or other medically important events), investigators report to the IRB within 24 hours of awareness. Suspected unexpected serious adverse reactions (SUSARs) are reported to the regulatory authority. Information on all serious adverse events is documented on the adverse event form together with additional follow-up information, and a copy of the report is retained with the participant's case report form. 17. Data Management and Confidentiality Participant data are collected in anonymized form, and each participant is assigned a unique identification code (a three-digit subject number beginning at 001). A subject list matching participant names with unique identification codes is kept confidentially under the management of the principal investigator. Case report forms are stored in a locked safe at the institution, and organized data are stored on access-restricted computers. CGM data are used only for research purposes and, after study completion, are stored and managed in de-identified form via a dedicated research account, separately from clinical care; they are not used for any non-research purpose. After study completion, all documents, including original case report forms, are retained for three years. If study results are published, the requirements of the International Committee of Medical Journal Editors (ICMJE) are followed, and participant identity is strictly kept confidential. Even if a participant withdraws consent at any point after providing consent for the provision and research use of personal information, study-related records are retained for three years in accordance with the Enforcement Rule of the Bioethics and Safety Act (Article 15), and personal information is destroyed in accordance with the Enforcement Decree of the Personal Information Protection Act (Article 16).