Page 58 - Glucose Monitoring Devices

56 CHAPTER 4 Consequences of SMBG systems inaccuracy

Financial outcomes
Short term
Cost of therapy (e.g., insulin, meter, and strips) and acute complications (e.g., SHE
and DKA); the opportunity cost of managing the disease.

Long term
Quality-adjusted life years (QALYs), cost of managing long-term clinical complica-
tions, overall healthcare system costs.

Requirements
Given the complexity of the task, it is useful to reflect on what the requirements are
for a method for assessing the effects of accuracy. It is our position that assessment
of the effects of accuracy need to take into account canonical elements of variability
in the metabolicebehavioredeviceeenvironment system. Some of these elements
include the following:

Physiological variability
A study should trace the effects of measurement errors across a representative
sample of metabolic variability across the population, but also within the patient
(e.g., intraday variability of insulin sensitivity). The same meter error may have
different effects for two different persons or two different effects for the same person
at two different times.
Behavioral variability
As argued earlier and shown in Ref. [42], accuracy effects can be attenuated or
magnified to a large degree by human behavior. Accuracy studies should incorporate
sufficient variability for each therapy mode and use case of a meter, for example,
MDI versus CSII patients, the meter used for calibration only or also for decision-
making.

Device and lot variability
As shown in Refs. [11,13,14], each device has characteristic meter errors. Errors
vary from lot to lot [10,27] and even from device to device. When attempting to
analyze a single device, sufficient intradevice variability should be incorporated.
Most studies available pool all meter data together and present an aggregate (popu-
lation) analysis. Performance variability across meters of the same brand is critical
for accuracy risk characterization. When possible, accuracy data from postmarket
evaluations should be included, in case manufacturing process variations have
effects not present during development [55].
Therapy modes
Meters are used in multiple therapy modes and use cases. In some cases, as the only
source of information, or to confirm measurements obtained by another device