Eleven algorithm options are proposed for screening of people living with HIV for TB that include the new and existing screening tools presented in this section (see Annex 3). (See 3.3 for an introduction and discussion of screening algorithms in general, including the definitions and implications of single, parallel, sequential positive and sequential negative screening algorithms.)
The algorithms focus on screening and referral to a diagnostic evaluation, including an mWRD test, although LF-LAM should be used where indicated to enhance early detection of TB (12). Each algorithm has a different sensitivity and specificity and therefore different potential for true-positive, true-negative, false-positive and false-negative results. The yields of TB patients and predictive values also depend on the prevalence of TB in the population being screened. For all algorithms, the risk of a false-positive diagnosis increases as the prevalence decreases; therefore, attention must be paid to diagnostic accuracy, particularly when the prevalence of TB in the screened population is low.
The algorithms, when combined with mWRD for diagnosis, have different costs and requirements in terms of human resources and health systems. Which algorithm is chosen for screening and diagnosis depends on the risk group, the prevalence of TB, the availability of resources and the feasibility of implementing the algorithm. The tables in Annex 4 show modelled estimates of the performance and outcomes of the screening algorithms described below, including the results of true- and falsepositive diagnosis for the entire algorithm, consisting of the screening test(s), followed by diagnostic evaluation with an mWRD.
Fig. A.3.1 W4SS single screening algorithm (page 76)
Fig. A.3.2 CRP single screening algorithm (page 77)
Fig. A.3.3 CXR single screening algorithm (page 78)
Fig. A.3.4 Parallel screening algorithm with W4SS and CRP (page 79)
Fig. A.3.5 Sequential positive screening algorithm with W4SS and CRP (page 80)
Fig. A.3.6 Sequential negative screening algorithm with W4SS and CRP (page 81)
Fig. A.3.7 Parallel screening algorithm with W4SS and CXR (page 82)
Fig. A.3.8 Sequential positive screening algorithm with W4SS and CXR (page 83)
Fig. A.3.9 Sequential negative screening algorithm with W4SS and CXR (page 84)
Fig. A.3.10 mWRD single screening algorithm for medical inpatients in settings with TB prevalence > 10% (page 85)
Fig. A.3.11 mWRD single screening algorithm for people living with HIV (page 86)