Table 2.
Performance of CHI3L1 alone and in combination for predicting mortality among children with severe malaria a
| Markers | Cut-point b | Sensitivity (%) | Specificity (%) | PLR c | NLR | PPV (%) d | NPV (%) |
|---|---|---|---|---|---|---|---|
| CHI3L1 |
≥179.1 ng/mL |
91.3 (73.2-97.6) |
67.5 (56.6-76.7) |
2.8 (2.0-3.9) |
0.13 (0.034-0.49) |
14.5 (5.0-30.3) |
99.2 (93.1-100) |
| CHI3L1, IP-10, sICAM-1 |
≥2 points |
91.3 (73.2-97.5) |
83.8 (74.2-90.3) |
5.6 (3.4-9.4) |
0.10 (0.028-0.39) |
25.4 (9.2-48.8) |
99.4 (94.4-100) |
| CHI3L1, Ang-2, IP-10 |
≥2 points |
100 (85.7-100) |
81.3 (71.3-88.3) |
5.3 (3.4-8.4) |
-- |
24.4 (9.4-46.0) |
100 (95.4-100) |
| CHI3L1, Ang-2, sICAM-1 | ≥2 points | 95.7 (79.0-99.2) | 81.3 (71.3-88.3) | 5.1 (3.2-8.1) | 0.054 (0.008-0.37) | 23.6 (8.6-45.7) | 99.7 (94.8-100) |
aAll parameters are presented with 95% CIs in parentheses. bCut-points were determined using the Youden Index (J = max[sensitivity + specificity – 1]). cPLR, positive likelihood ratio; NLR, negative likelihood ratio; PPV, positive predictive value; NPV, negative predictive value; sICAM-1, soluble ICAM-1, Ang-2, angiopoietin-2. dPPVs and NPVs were based on estimates that 5.7% of CM and SMA patients at Mulago hospital die of their malaria infection [34].