35 Continuous Glucose Monitoring Performs Comparably to Point-of-Care Glucose Testing in Hospitalized Pediatric Patients

Diabetes mellitus is a major global health issue, impacting more than 500 million people worldwide and over 38 million in the United States. Continuous glucose monitoring (CGM) has transformed diabetes management by offering real-time glucose readings through minimally invasive sensors, reducing the need for frequent fingersticks. While studies in adults have demonstrated the clinical utility of CGM in both outpatient and inpatient settings, evidence supporting its use in hospitalized pediatric patients is limited and no CGMs to-date have been cleared by the FDA for in-patient use. Our goal was to assess the accuracy and clinical impact of CGM use in hospitalized pediatric patients. We hypothesized that the clinical performance of CGM is comparable to standard of care (SOC) glucose testing. This retrospective study consisted of 72 unique hospitalized pediatric patients with dysglycemia. We analyzed paired glucose concentrations measured by CGM and collected within 10 minutes then tested by SOC testing using in-lab or point-of-care (POC) glucose methods. Performance was assessed by calculating the mean absolute relative difference (%MARD) and determining the percentage of glucose values in acceptable zones using Clarke Error Grid analysis. We also assessed the performance of CGM and POC in the hypoglycemic range and the frequency with which CGM was confirmed by SOC when testing in the hypoglycemic or hyperglycemic range as recommended by guidelines. Cohen’s kappa was calculated using 70 mg/dL and 180 mg/dL as the threshold for hypoglycemic and hyperglycemic. 41 males (56.9%) and 31 females (43.1%) were included in the study with a median age of 13.3 years (IQR: 7.2 to16.2). The median HbA1c at admission was 8.4% (IQR: 6.8 – 10.8), and the median BMI was 20.4 (IQR: 16.1 – 25.4). A total of 2,228 matched CGM and SOC glucose pairs were analyzed, yielding an overall MARD of 14.8%, with over 94% of values falling within the clinically acceptable Clarke Zones A and B. However, the MARD was 20.2% when the CGM result was <70 mg/dL with 38.6% of specimens exceeding the FDA threshold of +/- 15mg/dL and a bias of 1.38 mg/dL in POC values relative to CGM. The Cohen’s kappa coefficient for CGM and POC paired glucose concentrations was 0.66 (95% CI: 0.63 – 0.69). Notably, 125 (28.0%) CGM readings below 70 mg/dL were not rechecked within 30 minutes using SOC methods, and 1,473 (84.7%) readings above 200 mg/dL were not confirmed with SOC glucose tests. In conclusion, the performance of CGM in hospitalized pediatric patients is clinically acceptable but has a MARD that exceeds acceptable thresholds at low glucose concentrations, implying a need to confirm low readings by SOC methods. However, confirmatory testing is frequently not performed in clinical practice when CGM values are outside of normal limits.