Two episodes of severe hypoglycemia occurred while on the closed-loop system, one due to user error and the other for unknown reasons. Both patients made a full recovery, Dr. Thabit said.
The closed-loop system was interrupted unintentionally an average of once every 5 nights, which one physician in the audience called "very impressive."
The separate day-and-night study of 17 adults randomized them to 8 days of either therapy followed by a 1- to 3-week washout period and then 8 days on the other therapy. Again, patients received a day of training at a research facility before using the closed-loop system at home.
During the 7 days of home use with the closed-loop system, patients spent significantly less time in hyperglycemia (levels above 10 mmol/L), compared with the control therapy (21% vs. 30% of the time, respectively), reported Dr. Lalantha Leelarathna, also of the University of Cambridge. Time spent in hypoglycemia did not differ significantly between groups.
Mean glucose levels were significantly lower while patients were on the closed-loop system (8.1 mmol/L) than on the sensor-augmented pump (8.8 mmol/L), he said.
The closed-loop system was interrupted unintentionally every 12 hours on average, mainly because the pump was not connected, the continuous glucose monitor was unavailable, or the user changed pump settings. Two severe hypoglycemic episodes occurred, one during use of the closed-loop system and one during the crossover period. Both patients recovered fully. Four episodes of high glucose because of failure of the insulin infusion set did not lead to ketosis or hospitalization.
"We need to improve portability and connectivity between devices," Dr. Leelarathna said.
The Artificial Pancreas at Home (AP@home) consortium next will test the system in a 3-monthy study of day-and-night use by 30 adults with type 1 diabetes, compared with optimized sensor-augmented pump therapy, Dr. Leelarathna said.
Previous studies conducted exclusively in clinical research facilities have shown that closed-loop insulin delivery systems can improve glucose control, but these are among the first studies to show that they may do the same when used by patients at home.
Dr. Leelarathna and Dr. Thabit reported having no financial disclosures.
On Twitter @sherryboschert