Screening algorithms options are presented in section9. Each algorithm for adults includes options for the initial diagnostic testing of people whose screening test is positive: either sputum-smear microscopy (conventional light microscopy used to examine direct smears stained with Ziehl–Neelsen, with or without specific sputum-processing methods, or fluorescence microscopy including microscopy with light-emitting diodes) or a rapid molecular test that has been demonstrated to have high accuracy, such as the Xpert MTB/RIF test (Cepheid, Sunnyvale, CA) (or any rapid test recommended by WHO in the future that has the same or better accuracy than the Xpert MTB/RIF ).
Positive or negative diagnostic results may require a repeat test or further diagnostic evaluation using culture, drug-susceptibility testing, clinical assessment, or some combination of these. In these algorithms, culture is not considered for use as an initial diagnostic test because it requires a much longer wait for results (2–6 weeks) than both nucleic acid amplification tests (such as the Xpert MTB/RIF test) and sputum-smear microscopy, both of which can provide final test results in less than 1 day.
The choice of algorithm depends on the risk group being evaluated, the prevalence of TB, the availability of resources and the feasibility (see Sections 8 and 9).
For each algorithm, the following estimates are provided for different prevalences of TB (0.5%, 1% and 2%) in the screened population:
negative predictive values of the screening test;
pretest probability for the initial diagnostic test;
positive predictive value and negative predictive value of the initial diagnostic test among people whose screening test is positive;
proportion of true cases detected by the algorithm using outcomes from only the initial diagnostic test;
proportion of those with a negative result on the initial diagnostic test that is assumed to undergo further diagnosis with chest X-ray (if not already done) and clinical assessment (see support material at
www.who.int/tb/tbscreening);
proportion of true cases detected by the algorithm using the results of the initial diagnostic test plus clinical diagnosis in the proportion of those with a negative result on the diagnostic test that is assumed to undergo clinical diagnosis;
positive predictive value when using the combination of the initial diagnostic test and clinical diagnosis.
Definitions of estimates used to evaluate the algorithms
Positive predictive value (PPV): the likelihood that a person diagnosed with TB has true culture-positive TB (also the proportion of all detected cases that are true culture-positive TB cases)
Negative predictive value (NPV): the likelihood that a person who is not diagnosed with TB does not have culture-positive TB (1 – NPV = the probability that a person not diagnosed with TB actually has culture-positive TB)
Pretest probability (PTP): the prevalence of culture-positive TB among persons eligible for a test (for a second test in an algorithm this equals the PPV of the previous test); the pretest probability increases with each screening step
Flow charts of algorithms
(chest X-ray and Xpert MTB/RIF not available).
(chest X-ray not available, Xpert MTB/RIF available).
(chest X-ray available, Xpert MTB/RIF not available).
(chest X-ray and Xpert MTB/RIF available).
(chest X-ray and Xpert MTB/RIF not available).
(chest X-ray not available, Xpert MTB/RIF available).
(chest X-ray available, Xpert MTB/RIF not available).
(chest X-ray and Xpert MTB/RIF available).
(Xpert MTB/RIF not available).
(Xpert MTB/RIF available).