Table 1.
Characteristics of the study population
| UMass dataset | Study population | Events used for analysisb | |
|---|---|---|---|
| Total stays | 72,650 | 500 | 500 |
|
| |||
| Sex, male | 41,472 (57.1) | 280 (56.0) | - |
|
| |||
| Age (yr) | 64 (52–76) | 65 (53–76) | - |
|
| |||
| Ethnicity | - | ||
| White | 63,144 (86.9) | 426 (85.2) | |
| Non-White | 4,188 (5.8) | 33 (6.6) | |
| Unknown | 5,241 (7.2) | 41 (8.2) | |
|
| |||
| ICU type | - | ||
| Medical | 30,730 (42.4) | 239 (47.8) | |
| Surgical | 14,644 (20.2) | 86 (17.2) | |
| Cardiac | 13,402 (18.5) | 77 (15.4) | |
| Neuro | 13,770 (19.0) | 98 (19.6) | |
|
| |||
| LOS (hr) | 59 (34–110) | 114 (45–250) | - |
|
| |||
| Mortalitya | 8,613 (11.9) | 113 (22.6) | - |
|
| |||
| Vasopressora | 10,429 (14.4) | 240 (48.0) | 219 |
|
| |||
| Mechanical ventilationa | 7,386 (10.2) | 218 (43.6) | 146 |
Values are presented as number of patients (%) or median (25th–75th percentile).
LOS: length of stay in intensive care unit.
a
There is no statistically significant difference in the parameters compared between UMass dataset and our study population, except for the characteristics that were part of the case stratification process, which improves statistical efficiency while preserving randomization (p < 0.05). The diversity of the patient population helps mitigate concerns associated with model development based on data from a single center.
b
To avoid mis-tagging some events occurring during the first hours of the stay or very near to a previous event were excluded as explained in the method section.