Artificial pancreas treatment consists of an insulin pump, continuous glucose monitoring, and an algorithm to deliver insulin (and glucagon in dual-hormone systems) in response to glucose levels. In this meta-analysis published in BMJ, which included 40 randomised trials (1,042 participants), researchers compared artificial pancreas treatment with any type of insulin-based treatment in people with type 1 diabetes. Study duration was as long as 30 weeks.

The primary outcome — the proportion of time that the sensor glucose level was within the near-normoglycemic range (70–180 mg/dL) — was significantly higher with artificial pancreas treatment during 24 hours (140 additional minutes) and overnight than with control treatments. Proportions of time in hyperglycemia and hypoglycaemia, mean levels of sensor blood glucose during 24 hours, and glycosylated haemoglobin levels all were significantly lower with artificial pancreas treatment. Mean daily need for insulin did not differ between groups.