Introduction
Outcome prediction in the critically ill obstetric patient is controversial.
Objective
To determine the applicability of SAPS II, APACHE II and APACHE III-J scores calculated at day 1 of admission in outcome prediction.
Design
An open prospective data collection, as part of the Assessment of Prognosis and Risk of Mortality in Obstetrics (APRiMO) study. Discrimination was assessed by area under the receiver operator characteristic curve (AUROC). Calibration was assessed by Hosmer–Lemeshow (HL) C statistic. P < 0.05 was considered significant.
Setting
A multidisciplinary ICU.
Study period
January 1996–December 2003.
Patients
Obstetric patients (n = 541) admitted for at least 4 hours in the ICU.
Measurements and results
Mean age was 31.2 ± 5.9, mean term was 34.7 ± 4.5 weeks. Seventy percent of admissions were obstetric complications. The mortality rate was 10.4% (n = 57). The three scores discriminated (Fig. 1) and calibrated (Table 1) well. APACHE III was the only score to calibrate well using original mortality prediction equations (Table 2).
Figure 1.
Receiver operator characteristic curves of the severity scoring systems' admissions.
Table 1.
| Scoring system | Survivors | Non-survivors | AUROC | HL C statistics | SMR |
|---|---|---|---|---|---|
| SAPS II | 20.35 | 55.7 | 0.926 ± 0.04 | 0.399 | 0.96 |
| APACHE II | 7.07 | 21.2 | 0.918 ± 0.035 | 0.653 | 1.06 |
| APACHE III | 23.1 | 81.97 | 0.917 ± 0.045 | 0.73 | 1.43 |
Table 2.
| Scoring system | AUROC | HL C statistics |
|---|---|---|
| SAPS II | 0.925 | 0.036 |
| APACHE II | 0.785 | 0.02 |
| APACHE III | 0.922 | 0.15 |
Conclusion
The three scores are good discriminators. Customization of the mortality prediction equations seems necessary. APACHE III-J seems to be the best (adjustment by adequate diagnostic categories).