TABLE 1.
COMPARISON OF THE TRAINING DATA SET, VALIDATION DATA SET 1, AND VALIDATION DATA SET 2
| Training Data Set | Validation Data Set 1 | Validation Data Set 2 | |
|---|---|---|---|
| No. of patients | 40 | 139 | 193 |
| No. of deaths (%, 95% CI) | 8 (20, 10–36) | 18 (13, 8–20) | 26 (139–19) |
| Median age, yr (IQR) | 2.8 (1.3–3.9) | 2.2 (0.8–4.7) | 3.2 (2.0–5.9) |
| Median PRISM score (IQR) | 17 (11–22) | 16 (9–22) | 16 (12–21) |
| Sex, no. male/female | 27/13 | 85/54 | 109/84 |
| Ethnic origin: n (%) | |||
| White | 22 (55) | 91 (65) | 125 (64.8) |
| Hispanic | 3 (7) | 10 (7) | 44 (22.8)* |
| African descent | 11 (27) | 28 (20) | 8 (4.1)* |
| Asian | 2 (5) | 4 (3) | 6 (3.1) |
| Other (mixed racial) | 2 (5) | 6 (4) | 10 (4.7) |
| Median IL-8, pg/ml (IQR) | 97 (0–800) | 93 (0–501) | 336 (123–1,502)† |
| No. of patients with IL-8 > 220 pg/ml (% mortality, 95% CI) | 14 (43, 19–70)‡ | 57 (25, 14–38) | 121 (18, 12–26) |
Definition of abbreviations: CI = confidence interval; IQR = interquartile range; PRISM = Pediatric Risk of Mortality.
Differences are not significant unless otherwise specified.
*
P < 0.05 versus training data set and validation data set 1 (χ2).
†
P < 0.05 versus training data set and validation data set 1 (analysis of variance).
‡
P = 0.20 versus validation data set 1 (Fisher exact test), and P = 0.07 versus validation data set 2 (χ2 test).