Table 4. Comparison of Different Approaches to Prevent Tuberculosis in People Living With HIV.
| Study Algorithm*(N = 772) | IGRA-alone Approach* | WHO Approach* | ||
|---|---|---|---|---|
| Variable | pVL/CD4 approach† | pVL/CD4 approach augmented by IGRA** | (N = 772) | (N = 772) |
| Isoniazid preventive treatment needed | 228 (29.5) | 300 (38.9) | 90 (11.7) | 772 (100.0) |
| IGRA testing needed | 0 | 544 | 772 | 0 |
| HIV viral load needed | 772 | 772 | 0 | 0 |
| CD4 cell count needed | 772 | 772 | 0 | 0 |
| False-negative results | 8 (1.5) | 4 (0.8) | 11 (1.6) | 0 (0.0) |
| True positives (Sensitivity) | 9 (52.9) | 13 (76.5) | 6 (35.3) | 17 (100.0) |
| Patient unnecessarily treated | 219 (28.4) | 287 (31.2) | 84 (10.9) | 755 (97.8) |
| Number needed to treat (NNT) (95% CI)assuming 90% efficacy of IPT | 25 (14–69) | 23 (14–47) | 15 (7–81) | 45 (29–81) |
| Number needed to treat (NNT) (95% CI)assuming 50% efficacy of IPT | 51 (39–71) | 46 (37–61) | 30 (21–54) | 91 (79–108) |
| Characteristics of patients with true positive results selected for treatment | ||||
| No of patients | 9 | 13 | 6 | 17 |
| Median CD4 cell count at presentation (cells/μl) | 272 (39–441) | 287 (39–932) | 639 (76–932) | 365 (39–932) |
| Median months to onset of tuberculosis | 13.3 (2.8–42.5) | 24.5 (2.8–64.0) | 48.1 (2.8–64.0) | 26.3 (2.8–64.0) |
| Characteristics of patients with false-negative results | ||||
| No of patients | 8 | 4 | 10 | — |
| Median CD4 cell count at presentation (cells/μl) | 639 (365–932) | 462 (365–902) | 337 (39–902) | — |
| Median months to onset of tuberculosis | 46.5 (4.4–64.0) | 32.8 (4.4–49.6) | 24.5 (4.4–49.6) | — |
Data are described as n (%) or median (range), unless otherwise specified. CI, confidence interval. IGRA; interferon-gamma release assay; IQR, interquartile range; pVL, HIV plasma viral load; WHO, World Health Organization.
* pVL/CD4 approach include those with high HIV viral load of > = 100,000 copies/mL and low CD4 cell counts of < 350 cells/μL
**pVL/CD4 approach augmented by IGRA includes those in the pVL/CD4 approach and additionally those with a positive IGRA test; IGRA approach include those with positive IGRA tests only and the W.H.O. approach does no testing and includes all.
†In those with CD4 ≥ 350 cells/μL, the proportion testing IGRA-positive was 72/545 (13.2%), and in those with high HIV plasma viral load ≥ 100,000 copies/mL, 1/14 (7.1%).