Cryptococcal Meningitis Screening and Treatment

The CM model begins with the projected number of PLHIV‐initiating ART with CD4 <200 cells/mm³, drawn from 2021 national AIM files in Spectrum (Figure 2 and Table 3). We then calculate the number of cryptococcal deaths that would be expected in the absence of any CM screening/treatment intervention, based on the prevalence η of cryptococcal antigenemia (CrAg+), the fraction θ of CrAg+ who go on to develop CM and a CFR for CM. Values for these parameters were drawn from expert opinion and published studies [5 (link), 6 (link), 17 (link), 18 (link), 19 (link), 26 (link)]. Finally, we reduce these deaths based on the fraction F that are averted by intervention, where the value of F depends on whether AHD cases are identified by CD4 testing or clinical staging.
The fraction F of deaths averted by intervention is calculated by dividing CrAg+ AHD patients into three groups: those with symptomatic (we assume that multiple symptoms of CM would be detected in this group) CM at presentation (Group 1, p₁), those with sub‐clinical CSF‐positive cryptococcal disease (Group 2, p₂) and those with CSF‐negative cryptococcal antigenemia (CrAg+ but CSF–, Group 3, p₃), such that p1+p2+p3=1. The model assumes three treatment regimens, each associated with a coverage ei, and four treatment efficacies τi of preventing death from CM. The first treatment is an amphotericin‐based regimen, while the second is a high‐dose fluconazole regimen, used when amphotericin is unavailable (e2=1−e1). We assume that in many settings, patients with CM receive fluconazole monotherapy, despite it no longer being a WHO‐recommended regimen, and that coverage of amphotericin‐based treatments is relatively low. The third treatment is fluconazole‐based preventive therapy for CrAg‐positive patients who are CSF‐negative (efficacy τ₃), or who have sub‐clinical CM that is not detected by CSF CrAg testing (efficacy τ₄). This third regimen is available with independent coverage e₃.
We assume Group 1, those with symptomatic CM, are easily diagnosed and receive whichever of the two CM treatment regimens is available. Groups 2 and 3 must first be screened for AHD and correctly identified as such (α). If identified as AHD, Group 2 receive a test for cryptococcal antigenemia (coverage c₂), followed a test for CSF‐positivity (coverage c₁); a positive CSF test is followed by the available treatment regimen, while a negative CSF test is followed by pre‐emptive therapy. If identified as AHD, Group 3 receive a test for cryptococcal antigenemia followed by pre‐emptive therapy.
As with the TB model, the effect of CD4 testing on cryptococcal disease/CM mortality comes entirely through the identification of AHD, in this case for Groups 2 and 3. We assume that testing for both CrAg‐positivity and CSF‐positivity was indicated equally for all AHD cases, whether identified by clinical staging or CD4 testing. We assumed that CD4 testing would successfully identify all PLHIV with CD4 <200 cells/mm³ as having AHD. There are limitations of clinical staging for the identification of CD4 <200 cells/mm³; estimates of the sensitivity of clinical staging can vary but results from Munthali et al. showed that 60% of AHD cases (CD4 <200 cells/mm³) were identified by clinical staging [26 (link)], which we have adopted in our analysis.