Skip to main content
. 2022 Jul;3(7):e521–e532. doi: 10.1016/S2666-5247(22)00062-3

Figure 1.

Figure 1

Diagnostic yield of individual and combination rapid diagnostic tests as a function of clinical variables

Individual test (top row panels) and combinations of these tests (bottom row panels) diagnostic yield modelled as a function of three a priori specified patient variables: CD4 count (left column), haemoglobin (middle column), and venous lactate (right column). Plotted lines show a Locally Estimated Scatterplot Smoothing function fit to the proportion of tuberculosis cases identified by the diagnostic tests by different strata of the patient variables. Shaded areas indicate 95% CIs derived from 1000 bootstrapped resampling and refitting of the Locally Estimated Scatterplot Smoothing models. Xpert=Xpert MTB/RIF. Xpert Ultra=Xpert MTB/RIF Ultra. LAM=lipoarabinomannan.