Micro December 2018

<0.5. The largest differences between methods were for the stainless steel, -0.154 for the low 1

level and -0.174 for the high level, still well below the 0.5 criterion. Except for the stainless 2

steel, all CIs were within (-0.150, 0.150). For the stainless steel, CIs were (-0.414, 0.106) for the 3

low level and (-0.303, -0.044) for the high level, still within the (-0.5, 0.5) recommendation. For 4

the 3M Petrifilm Rapid E. coli /Coliform Count Plate compared to ISO 16649-2:2001 for E. coli 5

(Table 15), mean differences were ≤ 0.054 log 10 in all levels of all foods tested and the sealed 6

concrete. As with the total coliform results, the largest differences between methods were seen 7

in the stainless steel, -0.317 for the low level, -0.246 for the medium level, and -0.090 for the 8

high level. Except for the stainless steel, all CIs were within (-0.200, 0.200). Only the low level of 9

stainless steel (-0.648, 0.014) was outside of the recommended criterion. 10

11

In addition to the method comparison data, no significant differences were indicated in s r

between the 3M Petrifilm Rapid E. coli /Coliform Count Plate and any of the reference methods. 12

Standard deviations in most cases were <0.2 and did not show bias toward either method. 13

14

15

Robustness Study

16

17

This study evaluated the ability of the 3M Petrifilm Rapid E. coli /Coliform Count Plate to

remain unaffected by variations in method parameters that might be expected to occur when 18

the method is performed by an end user. The effects of perturbations in two method 19

parameters were investigated; sample volume and incubation time. Using a factorial design, 20

four treatment combinations were evaluated and compared to the nominal test condition: 1 ) 21

0.9 mL sample volume, 16 h incubation time; 2 ) 0.9 mL sample volume, 28 h incubation time; 3 ) 22

24