147 / 453
CONFIDENTIAL INFORMATION
Page 5 of 13
the candidate method’s “confirmation stage” (confirmed) results (POD
CC
) at each
level of inoculation in order to attain a difference in POD for the two stages
(dPOD
CP
) at each level of inoculation. A 95% confidence interval was also
calculated for each dPOD
CP
value that was calculated, and was used to determine
whether
the “screening stage” results is
significantly different from the
“confirmation stage” results for each level of inoculation. According to the
methodology, if the 95% confidence interval of a dPOD
CP
value does not contain
zero, then the difference between the POD
CP
and the POD
CC
of the candidate
method at that level of inoculation is statistically significant at the 5% level. In
addition to these analyses, the rate of true positives (samples with a “screening
stage” positive result and “confirmation stage” positive result), the rate of true
negatives (samples with a “screening stage” negative result and “confirmation
stage” negative result), the rate of false positives (samples with a “screening stage”
positive result and “confirmation stage” negative result), and the rate of false
negatives (samples with a “screening stage” negative result and “confirmation
stage” positive result)
were calculated. Additionally, PODs were calculated for the
reference method’s “confirmed” results (POD
R
) at each level of inoculation and
were compared to the PODs that were calculated for each level of inoculation for
the candidate method where the “screening stage” and “confirmation stage” results
were both positive for a particular sample (POD
C
). The difference in POD for the
two values at each level of inoculation were calculated (dPOD
C
) along with its
corresponding 95% confidence interval. As noted above, if the 95% confidence
interval of a dPOD
C
value does not contain zero, then the difference between the
POD
C
of the candidate method and the POD
R
of the reference method at that level
of inoculation is statistically significant at the 5% level.
3.0 RESULTS
The objective of this study was to validate the 3M
TM
MDA for the detection of
E.
coli
O157:H7 in frozen blueberries. The inoculum concentration of
E. coli
O157:H7 was determined from Most Probable Number analysis, along with the
95% confidence intervals, as shown in Table 1. These inoculation levels
targeted were chosen in order to meet the targeted POD values for the different
inoculation levels. The concentrations achieved through inoculation were found
to be 1.700 x 10
-2
MPN/g for the high inoculation level and 1.305 x 10
-2
MPN/g for
the low inoculation level, according to the MPN analysis (Table 1).
All data obtained for the candidate method testing of frozen blueberries,
‘screening stage’ and ‘confirmation stage’ (25 g frozen blueberry analyzed by
3M
TM
MDA) is provided in Table 2. All data obtained for the reference method
testing of frozen blueberries (25 g frozen blueberry analyzed by the FDA BAM
Ch. 4A) is provided in Table 3. The number of true positives (samples with a