Continuous Glucose Monitoring Hypoglycemia Alarms in Critical Care: Clinical Utility and Workflow Integration

Continuous glucose monitoring (CGM) has been associated with reductions in inpatient hypoglycemia; however, little is known about the performance and integration of CGM hypoglycemia alerts and alarms in the intensive care unit (ICU). We evaluated hypoglycemia signal frequency, clinical response, and contributing factors within an ICU implementation study.

This prespecified secondary analysis was conducted within a prospective hybrid effectiveness-implementation study of adults in a medical ICU. Hypoglycemia signals included low alerts (<100 mg/dL), urgent low soon alerts, and urgent low alarms (<55 mg/dL). Signal frequency, progression to hypoglycemia <70 mg/dL, nursing response times, and protocol fidelity were evaluated. Consecutive signals occurring within 15 min were clustered into signal events. Confirmatory point-of-care (POC) glucose response time was calculated from the initial signal to the first subsequent POC measurement. All POC-confirmed hypoglycemic events (<70 mg/dL) underwent structured multidisciplinary root cause review.

Among 100 participants, 60% experienced at least one hypoglycemia signal, generating 1060 signals (3.6 ± 6.8 signals/patient-day) that clustered into 378 signal events. Fifty-one percent of signal events remained between 80 and 100 mg/dL, without falling below the institutional treatment threshold (<80 mg/dL). Ninety-five percent of urgent low alarms were preceded by an urgent low soon alert, a mean of 10.0 ± 7.8 min earlier. Median confirmatory POC response time was 33 min (interquartile range 12-51) and was significantly faster for lower glucose values and validated CGM signals. All, but one hypoglycemic event, were preceded by a validated CGM hypoglycemia signal. Case review identified intravenous insulin, carbohydrate coverage, and enteral feed-related factors as the most common contributors to hypoglycemia.

CGM hypoglycemia signaling integrated into ICU workflows can provide advance warning before hypoglycemic events while maintaining a modest contribution to alarm burden. Signal characteristics influenced nursing response, supporting further optimization of inpatient CGM alarm thresholds and implementation strategies.

ClinicalTrials.gov NCT#05571397 https://clinicaltrials.gov/study/NCT05571397.