Evaluation of a Predictive Low-Glucose Management System In-Clinic

Background: Predictions based on continuous glucose monitoring (CGM) data are the basis for automatic suspension and resumption of insulin delivery by a predictive low-glucose management feature termed "suspend before low," which is part of the Medtronic MiniMed (R) 640G combined insulin pump and CGM system. This study assessed the safety and performance characteristics of the system in an in-clinic setting at eight sites. Materials and Methods: In-clinic standardized increases in basal insulin delivery rates were used to induce nocturnal hypoglycemia in subjects (14-75 years) with type 1 diabetes wearing the MiniMed 640G system. The "suspend before low" feature was set at 65 mg/dL, and as a result, the predictive algorithm suspended insulin delivery when the forecasted glucose was predicted to be <= 85 mg/dL in 30 min (a 20 mg/dL safety buffer). Reference plasma glucose values (Yellow Springs Instruments [YSI], Yellow Springs, OH) were used to establish hypoglycemia and were defined as >= 2 consecutive values <= 65 mg/dL. Results: Eighty subjects were screened. Among the 69 successful completers, 27 experienced a hypoglycemic event and 42 did not, a prevention rate of 60%. The mean (+/- standard deviation) YSI value at the time of pump suspension was 101 +/- 18.5 mg/dL, and the mean duration of the 68 "suspend before low" events was 105 +/- 27 min. At 120 min after the start of the pump suspension events, the mean YSI value was 102 +/- 34.6 mg/dL. Conclusion: The MiniMed 640G "suspend before low" feature prevented 60% of induced predicted hypoglycemic events without significant rebound hyperglycemia.