Skip to main content

Table 3 Clinical performance and programmatic impact of three diagnosis algorithms

From: Performance and impact of GeneXpert MTB/RIF® and Loopamp MTBC Detection Kit® assays on tuberculosis case detection in Madagascar

Algorithm TP TN FP FN Sensitivity Specificity PPV NPV Molecular testing Additional case notification Total case notification LR+ LR-
n n n n % (95% CI) % (95% CI) % (95% CI) % (95% CI) n (%) n n
Algorithm 1
 Smear (n = 514) 148 310 3 53 73.6 (67.1–79.3) 99.0 (97.1–99.8) 98.0 (94.1–99.6) 85.4 (81.4–88.7) 0 (0) N/A 151 76.8 0.3
Algorithm 2
 GeneXpert MTB/RIF® (n = 509) 174 300 8 27 86.6 (81.1–90.7) 97.4 (94.9–98.8) 95.6 (91.4–97.9) 91.7 (88.2–94.3) 509 (100.0) 31 182 33.3 0.1
 Loopamp MTBC Detection Kit® (n = 517) 170 311 5 31 84.6 (78.9–89.0) 98.4 (96.2–99.4) 97.1 (93.3–99.0) 90.9 (87.4–93.6) 517 (100.0) 24 175 53.5 0.2
Algorithm 3
 Smear - followed by GeneXpert MTB/RIF® (n = 506) 177 295 10 24 88.1 (82.8–91.9) 96.7 (94.0–98.3) 94.7 (90.3–97.2) 92.5 (89.0–94.9) 358 (70.1) 36 187 26.9 0.1
 Smear - followed by Loopamp MTBC Detection Kit® (n = 514) 177 307 6 24 88.1 (82.8–91.9) 97.5 (95.0–98.8) 96.7 (92.9–98.7) 92.8 (89.4–95.1) 366 (71.2) 32 183 45.9 0.1