Cost effectiveness of cryptococcal antigen screening as a strategy to prevent HIV-associated cryptococcal meningitis in South Africa

Abstract

Objectives: Cryptococcal meningitis (CM)-related mortality may be prevented by screening patients for sub-clinical cryptococcal antigenaemia (CRAG) at antiretroviral-therapy (ART) initiation and pre-emptively treating those testing positive. Prior to programmatic implementation in South Africa we performed a cost-effectiveness analysis of alternative preventive strategies for CM.

Design: Cost-effectiveness analysis.

Methods: Using South African data we modelled the cost-effectiveness of four strategies for patients with CD4 cell-counts <100 cells/µl starting ART 1) no screening or prophylaxis (standard of care), 2) universal primary fluconazole prophylaxis, 3) CRAG screening with fluconazole treatment if antigen-positive, 4) CRAG screening with lumbar puncture if antigen-positive and either amphotericin-B for those with CNS disease or fluconazole for those without. Analysis was limited to the first year of ART.

Results: The least costly strategy was CRAG screening followed by high-dose fluconazole treatment of all CRAG-positive individuals. This strategy dominated the standard of care at CRAG prevalence ≥0.6%. Although CRAG screening followed by lumbar puncture in all antigen-positive individuals was the most effective strategy clinically, the incremental benefit of LPs and amphotericin therapy for those with CNS disease was small and additional costs were large (US$158 versus US$51 per person year; incremental cost effectiveness ratio(ICER) US$889,267 per life year gained). Both CRAG screening strategies are less costly and more clinically effective than current practice. Primary prophylaxis is more effective than current practice, but relatively cost-ineffective (ICER US$20,495).

Conclusions: CRAG screening would be a cost-effective strategy to prevent CM-related mortality among patients initiating ART in South Africa. These findings provide further justification for programmatic implementation of CRAG screening.

Figure 1. Simplified markov model structure.

Transition probabilities are listed as C1, C2, S1 etc. Variable names, descriptions and values are derived from table 1, as follows: S1 - Proportion with subclinical CM at baseline, CD4<50 cells/µL = 0.13, CD4 50–100 cells/µL = 0.03. S2 - Of those with subclinical CM, Proportion with CSF infection = 0.5. P1 - Proportion starting ART with CD4<50 cells/µl = 0.5. C1 - Probability of developing CM, by baseline CD4 category and time on ART: CD4<50 cells/µl = 252 per 1000 patient years (pyo) up to 3 months on ART, 72 per 1000 pyo 4–6 months on ART, 36 per 1000 pyo 7–9 months on ART, 0 per 1000 pyo 10–12 months. CD4 50–100 cells/µl = 56 per 1000 pyo up to 3 months on ART, 16 per 1000 pyo 4–6 months on ART, 8 per 1000 pyo 7–9 months on ART, 0 per 1000 pyo 10–12 months. R1 - Relative risk of CM with low dose fluconazole prophylaxis = 0.21. R2 - Relative risk of CM for CRAG positive taking high dose fluconazole prophylaxis = 0.1. R3 - Relative risk of CM for CRAG positive with amphotericin for CSF positive patients, fluconazole for CSF negative patients = 0. D1 - Probability of dying of acute CM = 45% dead at 1 month with CM. D2 - Probability of dying of CM within 1 year = 55% dead at 12 months on ART.

Figure 2. Cost Breakdown.

The cost of each of the four strategies divided into screening costs, preventive treatment (or “prevention”) costs and treatment costs. Screening costs (black shading) include all costs associated with CRAG screening including the CRAG assay. Prevention costs (grey shading) include all costs associated with prevention including universal fluconazole in the primary prophylaxis strategy; and fluconazole pre-emptive treatment for CRAG positive patients, LPs, clinic visits and in-patient amphotericin for screened patients with CNS involvement in the screen and treat strategies. Treatment costs (cross-hatched shading) include all costs associated with treatment of CM in patients who develop clinical CM. Costs are as outlined in table 2, and expressed as mean cost per patient/year in the ART programme.

Figure 3. Sensitivity analysis by background antigen prevalence.

The results of one-way sensitivity analysis varying the background cryptococcal antigen prevalence in patients entering ART programmes with CD4 cell counts <100 cells/µL. The cost of current standard of care (no prevention, or status quo) is shown by the dotted line, and the cost of the CRAG screening with targeted treatment of CRAG positive individuals with high dose fluconazole (no LPs) is shown by the solid line. The screen and treat strategy dominated the standard of care at antigen prevalences of 0.6% and higher. The shaded area represents the range of baseline CRAG prevalence figures reported in patients with CD4 counts <100 cell/µL at ART programme entry. Costs are expressed as mean cost per patient/year in the ART programme.