Figure 17 LR scattergram for diagnosis of MTB infection by NAAT compared with culture in studies using either in-house NAAT or commercial Xpert NAAT
LR = likelihood ratio; NAAT = nucleic acid amplification testing
The summary LR estimate for in-house NAATs was close to the border between the upper and lower quadrants or just below it, whereas the summary value for commercial NAAT was clearly in the upper half of the graph. This indicated that a positive NAAT result using commercial NAAT could predict culture-positivity with greater confidence than using an in-house NAAT for all specimen types. The summary estimates were also mostly within the green band in the upper right quadrant of the graph, indicating that although a negative result was likely to indicate a negative culture result, it could not rule out culture-positivity. For sputum specimens the summary estimate for in-house NAAT was just within the upper left quadrant, indicating more confidence in the specimen also being culture-negative.
The SROC curve, which depicts the relative trade-off between true-positive and false-positive results, indicated that NAAT performs well in predicting culture positivity, with an AUC of 0.97 (95%CI 0.95, 0.98) for all specimen types, 0.96 (95%CI 0.94, 0.97) for sputum specimens and 0.89 (95%CI 0.86, 0.91) for non-sputum specimens. The SROC curve showed some threshold effect, suggesting that in-house NAAT was less specific than the commercial Xpert NAAT when compared with culture, especially in countries with a high incidence of TB and when testing non-sputum specimens (Figure 18).
Share with your friends: |