Abstract
Introduction: The vital prognosis at hospital discharge of patients admitted to Critical Care Units (CCU) because a head injury, has been estimated using the Glasgow Coma Scale (GCS). The prognosis of those patients admitted to a CCU, has been estimated using the general severity of illness systems, Acute Physiology and Chronic Health Evaluation II (APACHE II), Simplified Acute Physiology Score II (SAPS II) and Mortality Prediction Model II ( MPM II ).Hypothesis: 1º.- To demonstrate that system MPM II performs better the hospital mortality than GCS, APACHE II y SAPS II, in those patients admitted to CCU because a head injury.
2º.- To demonstrate that system MPM II is reproducibile, and then, its conclusions trustworthy.
Material and Method: 1º.- 401 patients consecutively admitted to CCU because a head injury, from 137 CCUs in 12 European and North American countries. The GCS score was obtained, and the APACHE II, SAPS II and MPM II0 at admission and MPM II24 at 24 hours, scores too. Logistic regression was used to convert the GCS into a probability of death. Performance of the general severity systems (APACHE II, SAPS II, MPM II) and GCS, in order to predict the hospital mortality was assessed by evaluating calibration and discriminationusing The probabilities of death.
2º.- From 2.332 patients consecutively admitted to 15 CCU in Catalonia and Balearic Islands, a 5% random sample (119 cases) was studied, in order to assess the agreement between two differents observers. One collected the data prospectively, and the other one retrospectively.
Results: 1º.- The MPM II system, both at admission and at 24 hours, calibrate better than GCS, APACHE II and SAPS II. C statistis was 18.4 using GCS, 68.2 using APACHE II and 30.3 using SAPS II. All of them showed statistic differences between predicted mortality and observed results (p<0.05). MPM II0 and MPM II24 showed a lower C statistic ( 11.8 and 10.8 respectively), and did not showed ststistic differences between predicted and observed mortality (p>0.30). The discrimination of all systems was very good, more than 89%.
2º.- The MPM II variables reproducibility, measured using the concordance kappa index, was moderate to high. The global reproducibility, measured using the intraclass correlation coefficient, was higher than 85%. Thus, the mortality predictions using MPM II are trustworthy.
Conclusions: The MPM II performs better the hospital mortality prediction in patients admitted to CCU because a head injury, than those calculated using GCS, APACHE II or SAPS II. The results obtained using MPM II are, in group of patients, trustworthy and reproducibles.
| Date of Award | 20 Nov 2001 |
|---|---|
| Original language | Spanish |
| Supervisor | Lluis Marruecos Sant (Director) & Àlvar Net Castel (Director) |