Page 331 - Glucose Monitoring Devices

338 CHAPTER 16 The dawn of automated insulin delivery

to suggest 1 week of manual mode use before transitioning to hybrid closed-loop
insulin delivery. Meal announcements are required and, for this, the user must enter
discrete carbohydrate amounts. An analysis of a subset of participants in the
pivotal trial led researchers to suggest increasing the insulin to carbohydrate ratio
by 10%e20% for optimal results [104].

Auto mode exits: what are they and why do they happen
Although users can choose to switch to and from auto mode, forced system exits
may occur for a variety of reasons. Periods of prolonged hyperglycemia (sensor
glucose >300 mg/dL for over 1 h or > 250 mg/dL for over 3 h) or maximum
algorithm-derived insulin delivery for 4 h suggest that there may be a reason the
user needs to reevaluate the hyperglycemia; thus, the system will exit auto mode.
Similarly, minimum algorithm-derived insulin delivery for 2.5 h also leads to an
auto mode exit. Finally, if there is no sensor glucose reading due to a missed calibra-
tion or the sensor is 35% off from fingerstick glucose, an exit may occur. In some
instances, the system will revert to “Safe Basal,” which is an algorithm-derived basal
rate that can last for up to 90 min. After that time, if the issue is not resolved the
system reverts back to manual mode. These exits were created for patient safety,
but some have found these exits nuisances, which may cause users to discontinue
automated insulin delivery.

The highs and lows of real-world use of the 670G system
Reports of clinical use of the 670G following its commercial launch have yielded
somewhat conflicting results. Stone et al. published a retrospective analysis of 3141
patients who uploaded their data to Carelink during the first 10 months following
the commercial launch [105]. Improvements in time spent in target range
(70e180 mg/dL)wereseenwithautomode whencomparedtomanualmode(73%
vs. 66%) across all age groups, though there was a reduction in time spent in both
hypoglycemic (<70 mg/dL 2.1% v 2.7%) and hyperglycemic (>180 mg/dL 24.6%
v 31.4%) ranges [105]. To date, the 670G system is currently used by more than
180,000 people. An analysis of more than 8-million patient-days of data showed
that users on average had 71% time in target range in all age groups [106].
Yet, work from clinics implementing this therapy has painted a somewhat
different picture. Indeed, data from the Barbra Davis Center following 51 youth
who started the use of the system found that after 6 months, 37% of their cohort
had discontinued automated insulin delivery [107]. Furthermore, youth who
continued use of the system was found to have waning sensor use over time (81%
vs. 71%) as well as a decrease in time in target range (61% vs. 56%) [107]. These
findings have been echoed by similar observational studies conducted at both Boston
Children’s Hospital and Lucile Packard Children’s Hospital, where discontinuation
rates were 19% and 50%, respectively [108,109]. Importantly, both of these groups
reported that auto mode use led to a lowering in the HbA1c level [108,109].