Table 1.
Model Input Parameters
| South Africa | Malawi | Deterministic Sensitivity Analysis Range | References | |
|---|---|---|---|---|
| Cohort characteristics | ||||
| Age, median [IQR], years | 37 [30–46] | 38 [32–47] | … | [5] |
| Men/women, % | 50/50 | 37/63 | … | [5] |
| CD4 count at admission, median [IQR], cells/µL | 236 [70–445] | 219 [86–431] | … | [5] |
| TB prevalence,a % | 29 | 24 | 15–45b | [5, 9, 27, 28] |
| MDR-TB prevalence among those with TB, % | 3 | 1 | 1–7 (South Africa); 0.5–5 (Malawi) | [5, 29] |
| Patients able to provide sputum, % | 50 | 50 | 30–90b | Assumption [5, 6, 25] |
| Probability of empiric treatment, Xpert,c % | 11 | 11 | 0–20b | [5, 30] |
| Probability of empiric treatment, Xpert+AlereLAM and Xpert+FujiLAM, % | 10 | 10 | 0–20b | [5, 6] |
| Loss to follow-up from TB care after hospital discharge, %/month | 3.6 | 3.6 | 50–200% of base case valueb | [31, 32] |
| Mortality | ||||
| Death from untreated TB, monthly probability | 0.086 | 0.086 | 25–200% of base case valueb | [33, 34] |
| Death from AIDS (besides TB), CD4-dependent, monthly probability | 6.2 × 10–5–0.2 | 6.2 × 10–5–0.2 | … | [35, 36] |
| Cost of treatmentd | ||||
| DS-TB treatment cost, monthly (6-month duration), USD | $7 | $7 | … | [37] |
| MDR-TB treatment cost, monthly (24-month duration), USD | $231 | $231 | … | [37] |
| First-line ART costs (TDF/3TC/EFV), monthly, USD | $11 | $11 | 50–75% of base case value | [38] |
| Cost of TB diagnostic assay, per-test (USD) | ||||
| Sputum Xperte | $15 | $25 | … | [19, 20] |
| Urine AlereLAM | $3 | $3 | … | [21] |
| Urine FujiLAM | $6 | $6 | $3–20 | Estimate |
| Sensitivity | Specificity | |||
| Performance characteristics of diagnostic assays and strategies | ||||
| Diagnostic assayf | ||||
| Sputum Xpert, CD4 <200/≥200 cells/µL | 65%/65% | 98%/98% | … | [6], Assumption |
| Urine AlereLAM CD4 <200/≥200 cells/µL | 48%/2% | 97%/99% | … | [6], Assumption |
| Urine FujiLAM, CD4 <200/≥200 cells/µL | 62%/23% | 94%/98% | Sensitivity: 48%/8% to 77%/38%; specificity: 75–90% | [6], Assumption |
| Xpert Ultra, CD4 <200/≥200 cells/µL | 77%/77% | 96%/96% | … | [23] |
| Diagnostic Yield | ||||
| Diagnostic strategyf | ||||
| Xpert, CD4 <200/≥200 cells/µL | 33%/33% | … | [6], Assumption | |
| Xpert+AlereLAM, CD4 <200/≥200 cells/µL | 62%/35% | −20% to +20% of base case value | [6], Assumption | |
| Xpert+FujiLAM, CD4 <200/≥200 cells/µL | 70%/47% | −20% to +20% of base case value | [6], Assumption |
Abbreviations: ART, antiretroviral therapy; DS, drug-susceptible; EFV, efavirenz; HIV, human immunodeficiency virus; IQR, interquartile range; LAM, lipoarabinomannan; MDR, multidrug-resistant; TB, tuberculosis; TDF, tenofovir; USD, 2017 US dollars; 3TC, lamivudine.
aTB prevalence is the true prevalence among the simulated group of hospitalized patients with HIV.
bThese parameters were also examined in probabilistic sensitivity analysis using beta distributions (Supplementary Text).
cThose who were diagnosed clinically without microbiologic confirmation were empirically treated in the first month of model simulation.
dWe assumed that costs of TB drugs and ART drugs were equal across countries because they are imported across countries. Costs shown here are for drugs only.
eXpert cost in a Malawi-specific costing study was higher than the cost reported in South African studies and by the South Africa National Health Laboratory Service [19]. This is due to factors such as different costs of maintenance and repair and different economies of scale.
fThe indicated sensitivity of each assay is the sensitivity among those who provided a specimen and is independent of other test results. Italics reflect a diagnostic strategy rather than a single test. The diagnostic strategy yields applied in the model accounted for nonprovision of sputum specimens and for concordance between test results—eg, adding FujiLAM would increase diagnostic yield only if FujiLAM detected additional TB cases not detected by Xpert. In multitest strategies, we applied the lowest specificity of any individual test.