While 10mmol/L is the upper limit of normal BG levels, this may in practice indicate

that levels are much higher. Together, this information about glucose control reveals that, while convenient, pump therapy might be less effective than reported, although not necessarily less effective than MDI therapy. It may be that an anonymised survey elicits information that differs from other sources for a variety of reasons that relate to surveys in general as well as to diabetes. It also implies that despite being on a reliably constant basal dose of insulin and with boosts conveniently selected for delivery to a tailored pattern coupled with features such as electronic memory and safety lockout features, respondents were commonly above the target BG range. An increase in BG with CSII may result from an occlusion of the BYL719 infusion line or cannula, although more commonly problems arise from human SGI-1776 chemical structure error, for example inaccurate carbohydrate estimation, inaccurate insulin carbohydrate

ratios, insulin sensitivity factors, as well as lifestyle factors such as exercise and stress. Whether the postprandial BG peak would be detected would depend on the user testing at the relevant times. The positive attitude towards an artificial pancreas such as INSmart focused on the control of BG and user independence as well as improved quality of life. Negative responses were perceptions about relying on an automated system that could possibly fail or not be reliable. The concept of an implantable device rather than an external (and therefore easily-removable) pump

was clearly worrying to some. There were comments about the need for comfort, the safety of implantation and maintenance including refill which would all need to be demonstrated for an INSmart type device to secure approval from the Medical Devices Directive in the UK25 (FDA in the USA). The behaviour, PIK3C2G attitude and use of existing external pump users from the open ended questions from this survey provided some useful feedback toward a redesign of the existing device which has now successfully been implanted into diabetic pigs. It is apparent that current external pumps have shortcomings which an implantable INSmart device could overcome: Automated delivery of insulin to real time changing glucose levels by the fast uptake of glucose in the peritoneum. No changing of infusion lines, rotation of sites and not visible. No moving parts or electronic power requirements. No need to regularly check BG levels. No need to bolus for meal times. However, an implantable INSmart device would still need to overcome risks such as leakage of insulin or smart gel, infection and surgery. The general consensus from the survey was that most respondents felt that an implantable artificial pancreas would be a close match to a functioning healthy pancreas and therefore appealing.