Skip to main content

Table 2 Sub-analyses of sensitivity and specificity by sample type and smear status

From: Diagnostic accuracy of the Xpert MTB/RIF assay for extrapulmonary and pulmonary tuberculosis when testing non-respiratory samples: a systematic review

Sample

Statistical model for pooled estimates

Number of studies

Sensitivity

Specificity

Pooled estimate (95% CI)

Median (IQR)

Pooled estimate (95% CI)

Median (IQR)

CSF

Cannot calculate

10

 

0.85 (0.75 – 1.00)

 

1.00 (0.98 – 1.00)

Pleural fluid

Bivariate

9

0.34 (0.24 – 0.44)

0.37 (0.27 – 0.72)

0.98 (0.96 – 0.99)

1.00 (0.98 – 1.00)

Non-pleural serous fluid

Cannot calculate

4

 

0.67 (Range 0.00 – 1.00)

 

1.00 (Range 1.00 – 1.00)

All tissue

Bivariate

12

0.88 (0.77 – 0.95)

0.90 (0.73 – 0.99)

0.98 (0.87 – 0.99)

1.00 (0.89 – 1.00)

Lymph node biopsy/FNA

Bivariate

7

0.96 (0.72 – 0.99)

0.97 (0.71 – 1.00)

0.93 (0.70 – 0.99)

1.00 (0.94 – 1.00)

Gastric aspirate

Bivariate

8

0.78 (0.69 – 0.86)

0.85 (0.74 – 1.00)

0.99 (0.98 – 0.99)

1.00 (0.99 – 1.00)

Smear positive

Random effects

9

0.95 (0.91 – 1.00)

1.00 (0.98 – 1.00)

  

Smear negative

Random effects

10

0.69 (0.60 – 0.80)

0.69 (0.61 – 0.83)

 Â