Page 105 - Glucose Monitoring Devices

104 CHAPTER 5 Modeling the SMBG measurement error

The BCN error model is implemented within the Device for glucose monitoring
model (block B in Fig. 5.12) to simulate SMBG measurements starting from the BG
value returned by the UVA/Padova T1D simulator, whenever the patient’s behavior
and treatment decisions model (block C in Fig. 5.12) requests a BG check by SMBG.
Specifically, if the simulated BG value, BG sim , is in the zone 1 of the SMBG error
PDF model (i.e., below 115 mg/dL), the SMBG absolute error, err abs , is sampled
from the PDF model identified in zone 1. Then, the simulated SMBG measurements,
SMBG sim , is obtained as follows:

SMBG sim ¼ err abs þ BG sim (5.11)
Conversely, if BG sim is in zone 2, the SMBG relative error, err rel , is sampled from
the PDF model identified in zone 2. Then, SMBG sim is obtained as follows:

BG sim
SMBG sim ¼ err rel $ þ BG sim (5.12)
100
The T1D patient decision simulator was preliminary used to compare nonadjunc-
tive CGM use, that is, the use of CGM to make treatment decisions, versus conven-
tional therapy based on SMBG in 20 virtual adults [51]. Results of standard outcome
metrics, for example, time in hypoglycemia, time in the target range, time in hyper-
glycemia, and the number of hypo/hyperglycemic events, calculated for both CGM
and SMBG scenarios, supported the noninferiority of CGM nonadjunctive use
versus SMBG use [51].
In 2016, the T1D patient decision simulator was used in collaboration with Dex-
com Inc. (San Diego, CA) to run simulations to assess the safety and effectiveness of
the nonadjunctive use of the Dexcom G5 Mobile sensor compared to conventional
SMBG therapy [52]. In particular, a 2-week ISCT was performed on 40,000 virtual
unique adults and pediatric patients, each defined by a different combination of
physiology and therapy parameters, showing that, in both the adult and pediatric
populations, the risk of hypo and hyperglycemia using CGM for insulin dosing
was equivalent to, or even lower than, that obtained using SMBG. These results
were included among all the material (clinical study data, analysis of human factors,
opinion of clinicians and experts, and the testimony of CGM sensor users) that Dex-
com Inc. presented to the Clinical Chemistry and Clinical Toxicology Devices Panel
of the US Food and Drug Administration (FDA)’s Medical Device Advisory
Committee on July 21, 2016 to ask for a change of label for the G5 Mobile device
from adjunctive to nonadjunctive use [53,54]. The panel voted in favor of the safety
and effectiveness of the Dexcom G5 Mobile nonadjunctive use, and 6 months later,
in December 2016, the FDA approved the Dexcom G5 Mobile as the first CGM
device for nonadjunctive use in the United States. This approval had a positive
impact also on CGM reimbursement criteria, as in March 2017 Medicare announced
covering the costs of therapeutic CGM devices, that is, CGM sensors approved for
nonadjunctive use, for all patients with T1D and T2D on intensive insulin therapy
[55]. These changes potentially will contribute to extend CGM use and stimulate
the development of new CGM-based applications for personalized diabetes treat-
ment and prevention [56].