18 Appendix B: Specications (con’t)
Accuracy
Table 1 below shows the distribution of all the data from the In-Clinic study on the Clarke Error Grid. Accuracy
was assessed by comparing the dierences between the FreeStyle Navigator system and the YSI laboratory reference.
e Clarke Error Grid Analysis evaluates the clinical relevance of the dierences by dividing a correlation plot
(Figure 1) into ve zones as described in Table 1. e YSI results and the corresponding glucose results from the
FreeStyle Navigator system together (called a ‘data pair’ or ‘matched data points’) determine what zone of the error
grid the results fall into. Table 1 also shows that glucose data measured by FreeStyle Navigator system on the arm
and on the abdomen have similar distribution on the error grid. is demonstrates that there is no dierence in the
performance of the system when worn on the abdomen or on the back of the upper arm.
Table 1. Clarke Error Grid Analysis. Continuous glucose results from FreeStyle Navigator System (mg/dL)
vs. the YSI (mg/dL)
Zone in the Clarke
Error Grid N (pairs of data) Percentage in the
dierent zones (%)
Sensor Insertion Site
Abdomen % Arm %
A16627 81.7 81.5 81.8
B3398 16.7 16.8 16.6
C19 0.1 0.1 0.1
D316 1.6 1.6 1.5
E2 0.0 0.0 0.0
Total 20362
ere is no dierence between the performance of the
system on the arm and the performance of the system on
the abdomen.
Error Grid Explanation
A – Clinically accurate; would lead to correct and safe treatment
B – Benign; would lead to benign decisions or no treatment decisions
C – Overcorrection; would lead to over-correction of normal glucose levels
D – Failure to detect; would lead to failure to detect and treat high or low glucose levels
E – Erroneous readings; would lead to erroneous treatment decisions
Customer Care: 1-866-597-5520

154 Section 18

Appendix B: Specifications