Table 3. Predictive performance of different multivariate models in multiple logistic regression.
| Models | Variation points | Score cut-off point | Sensitivity (%) (95% CI) | Specificity (%) (95% CI) | Area under ROC Curve (%) (95% CI) | Positive predictive value (%) (95% CI) |
| 1. Clinical-laboratory* | (−1/4) | ≥3 | 81.4 (74.0–88.0) | 65.2 (52.4–76.5) | 79.4 (72.0–87.0) | 82.0 (74.3–88.3) |
| 2. Clinical-laboratory* | 91.5 (82.5–97.0) | 53.0 (35.5–70.0) | 74.0 (61.3–86.3) | 79.3 (69.0–87.4) | ||
| 1. Clinical-laboratory*+L. chagasi-ELISA | (−1/8) | ≥5 | 91.5 (85.3–96.0) | 80.3 (69.0–89.1) | 93.1 (89.5–97.0) | 90.1 (84.0–95.0) |
| 2. Clinical-laboratory*+L. chagasi-ELISA | 90.1 (81.0–96.0) | 89.0 (74.0–97.0) | 91.0 (85.0–97.3) | 94.1 (86.0–98.4) | ||
| 1. Clinical-laboratory*+IFAT | (−1/7) | ≥4 | 90.0 (83.4–94.5) | 77.3 (65.3–87.0) | 90.4 (86.0–95.0) | 88.5 (82.0–93.4) |
| 2. Clinical-laboratory*+IFAT | 99.0 (92.4–100) | 78.0 (61.0–90.0) | 95.0 (89.0–100) | 90.0 (81.0–95.5) | ||
| 1. Clinical-laboratory*+rK39-ELISA | (−1/10) | ≥7 | 98.0 (93.4–99.5) | 88.0 (77.5–95.0) | 97.0 (94.1–100) | 94.0 (89.0–97.4) |
| 2. Clinical-laboratory*+rK39-ELISA | 96.0 (88.1–99.1) | 89.0 (74.0–97.0) | 93.4 (87.0–100) | 94.4 (86.4–98.5) | ||
| 1. Clinical-laboratory*+DAT | (−1/11) | ≥5 | 90.0 (83.4–94.5) | 97.0 (89.5–100) | 97.3 (95.4–99.2) | 98.3 (94.0–100) |
| 2. Clinical-laboratory*+DAT | 91.5 (82.5–97.0) | 92.0 (77.5–98.2) | 97.0 (88.0–99.1) | 94.0 (87.3–100) | ||
| 1. Clinical-laboratory*+rK39 rapid test | (−1/9) | ≥5 | 94.0 (88.1–97.3) | 95.5 (87.3–99.1) | 98.5 (97.2–100) | 98.0 (93.1–99.5) |
| 2. Clinical-laboratory*+rK39 rapid test | 90.1 (81.0–96.0) | 97.2 (85.5–100) | 95.5 (91.4–99.4) | 98.5 (92.0–100) |
1 Test sample; 2 Validation sample;
*The model Clinical-laboratory was composed by variables: Splenomegaly and Leukopenia. Points assigned to variables in the models: Cough = −1, leukopenia = 1, splenomegaly and IFAT = 3, L. chagasi-ELISA = 4, rK39 rapid test = 5, rK39-ELISA = 6 and DAT = 7.