TABLE 2.
Test method (replicates) | CC/BP(μg/ml) | Categorical agreement (%) |
Kappa | Sensitivity (%) [95% CI] |
Specificity (%) [95% CI] |
% of isolates with very major errors (no. of isolates with very major errors/total no. of isolates) |
% of isolates with major errors (no. of isolates with major errors/total no. of isolates) |
---|---|---|---|---|---|---|---|
Agar proportion (n = 577) | 0.25 | 96.7 | 0.9112 | 88.0 [81.7–92.7] | 99.8 [98.7–100.0] | 12.0 (18/150) | 0.2 (1/427) |
0.5 | 88.6 | 0.6532 | 56.0 [47.7–64.1] | 100.0 [99.1–100.0] | 44.0 (66/150) | 0 | |
1 | 82.0 | 0.3956 | 30.7 [23.4–38.7] | 100.0 [99.1–100.0] | 69.3 (104/150) | 0 | |
BMD frozen (n = 572) | 0.12 | 98.1 | 0.9484 | 97.8 [93.8–99.6] | 98.2 [96.4–99.2] | 2.2 (3/139) | 1.8 (8/433) |
0.25 | 97.6 | 0.9315 | 90.6 [84.5–94.9] | 99.8 [98.7–100.0] | 9.35 (13/139) | 0.2 (1/433) | |
BMD dry (n = 584) | 0.12 | 99.3 | 0.9822 | 100.0 [97.6–100.0] | 99.1 [97.7–99.7] | 0 | 0.9 (4/434) |
0.25 | 97.3 | 0.9256 | 89.3 [83.3–93.8] | 100.0 [99.2–100.0] | 10.7 (16/150) | 0 | |
MGIT960 (n = 594) | 1 | 99.8 | 0.9956 | 100.0 [97.6–100.0] | 99.8 [98.8–100.0] | 0 | 0.2 (1/444) |
2 | 93.3 | 0.8043 | 73.3 [65.5–80.2] | 100.0 [99.2–100.0] | 26.7 (40/150) | 0 |
CC data (indicated in micrograms per milliliter) apply to the agar proportion and MGIT960 methods, and MIC breakpoint data (indicated in micrograms per milliliter) apply to both the frozen and dry microtiter plates using the BMD method. Sensitivity data represent percentages of isolates classified as resistant by the test method against the total number of true resistant isolates. Specificity data represent percentages of isolates classified as susceptible against the total number of true susceptible isolates. Major error, resistant by the evaluated method and susceptible by the reference standard. Very major error, susceptible by the evaluated method and resistant by the reference standard. None of the errors occurred in all three replicates. Rows in bold indicate the concentrations at which the categorical agreement between the observed results and the expected results as well as the sensitivity/specificity at detecting an isolate as susceptible or resistant were highest. BMD, broth microdilution; BP, breakpoint; CC, critical concentration; CI, confidence interval; DST, drug susceptibility testing; MGIT, Mycobacteria Growth Indicator Tube.