3.2 Pretest probability and test accuracy considerations

The predictive values of a test vary depending on the prevalence of TB in the patient population being tested. Table 3.1 provides examples of population-level projections of the results of testing with the various mWRDs in settings with different TB prevalence, based on pooled sensitivity and specificity estimates that were extracted from the WHO diagnostic policy statements for each test. Tables 3.2-3.4 provide those same parameters for detection of resistance to RIF, INH and FQs, respectively. The sensitivity of the test may be lower when used for active case finding in a population screening context because such patients would be less sick and have a lower bacillary burden. In choosing a test to implement, countries will need to consider the possible trade-offs between higher or lower sensitivity and higher or lower specificity, based on the prevalence of TB in their country. False negative results may lead to missed opportunities to treat TB. False positive results may lead to the overtreatment of patients without TB. In some settings, countries may need to conduct additional modelling work to support decisions on implementation strategies, based on the trade-offs between sensitivity and specificity in their settings.

Usually, a decision to undertake a diagnostic work-up of an individual for TB begins with assessing symptoms and signs of TB disease. However, many individuals with culture-positive TB may not have symptoms or may consider the symptoms too insignificant to report, leading to missed diagnostic opportunities. To improve TB case detection and identify individuals suitable for TB preventive treatment, WHO has updated the TB screening guidelines.⁴⁴ Several modalities are recommended for screening: symptom screening, chest X-ray and mWRDs. For screening People with HIV, recommended screening modalities include the classical four-symptom screen, chest X-ray, and an mWRD or positive C-reactive protein test (>5 mg/L).

Chest X-ray as a screening or triage tool can identify individuals to be tested with an initial molecular test and can thus reduce the number of individuals tested and the associated costs but the CXR cost would need to be lower than the test costs (37, 38).⁴⁵ This approach is likely to improve the pretest probability for TB and therefore should improve the predictive value of the molecular test and reduce false positive results, especially in populations with a low prevalence of TB. For example, the addition of chest X-ray as a screening tool to an algorithm in which all individuals with an abnormal chest X-ray receive mWRD was calculated to increase the positive predictive value of the mWRD from 56.8% to 78.5% and the prevalent cases detected from 69% to 80%, compared with testing with an mWRD, irrespective of symptoms, in a population with a TB prevalence of 1% (39).⁴⁶

Performance of mWRDs for the detection of TB in adults

Performance of molecular tests for the detection of rifampicin resistance

 Performance of molecular tests for the detection of isoniazid resistance

Performance of molecular tests for the detection of fluoroquinolone resistance

⁴⁴ See https://www.who.int/publications/i/item/9789240022676

⁴⁵ See https://www.who.int/publications/i/item/9789241548601 and https://www.who.int/publications/i/item/9789241511506

⁴⁶ See https://www.who.int/publications/i/item/9789240022676

CI: confidence interval; CrI: credible interval; FN: false negative; FP: false positive; MC-aNAAT: moderate complexity automated nucleic acid amplification test; mWRD: molecular WHO-recommended rapid diagnostic test; Se: sensitivity; Sp: specificity; TB: tuberculosis; TN: true negative; TP: true positive.

ᵃ When used in a microscopy laboratory. When tested in reference laboratories, the sensitivities of Truenat MTB and Truenat MTB Plus were 0.84 and 0.87, respectively, and specificities were 0.97 and 0.95, respectively

CI: confidence interval; FL-LPA: line-probe assay for first-line drugs; FN: false negative; FP: false positive; MC-aNAAT: moderate complexity automated nucleic acid amplification test; Se: sensitivity; Sp: specificity; SS+: sputum-smear-positive; TB: tuberculosis; TN: true negative; TP: true positive.

ᵃ The rifampicin resistance detection by Xpert MTB/RIF, Ultra, Truenat MTB-RIF Dx and MC-aNAAT occurs only in case TB is detected; that is why suggested prevalence reflects rifampicin resistance in newly detected TB patients.

CI: confidence interval; FL-LPA: line-probe assay for first-line drugs; FN: false negative; FP: false positive; INH: isoniazid; LC/MC-aNAAT: low/moderate complexity automated nucleic acid amplification test; SS+: sputum-smear-positive; TB: tuberculosis; TN: true negative; TP: true positive.

ᵃ The isoniazid resistance detection by MC-aNAAT occurs only in cases where TB is detected. That is why suggested prevalence, reflecting INH resistance in newly detected TB patients, also applies to this technology class.

CI: confidence interval; FN: false negative; FP: false positive; LC-aNAAT: low complexity automated nucleic acid amplification test; SL-LPA: line-probe assay for second-line drugs; SS+: sputum-smearpositive; TB: tuberculosis; TN: true negative; TP: true positive.

Book navigation