Comparison of risk prediction models for admissions to UK critical care units following traumatic brain injury

Introduction

Traumatic brain injury is a potentially fatal condition and a common cause of admission to ICUs in the United Kingdom. We describe the case mix and outcomes of head injury patients requiring intensive care and evaluate the ability of five risk prediction models to predict mortality in these patients.

Method

Secondary analysis of a high-quality clinical database, the Intensive Care National Audit and Research Centre Case Mix Programme Database, on 374,594 admissions to 171 critical care units in England, Wales and Northern Ireland from 1995 to 2005. Calibration of risk prediction models was assessed by the area under the receiver-operating characteristic curve (AUROC), discrimination by the Hosmer-Lemeshow (H-L) C statistic and the intercept and slope from Cox's calibration regression, and overall fit by Brier's score.

Results

A total of 11,021 admissions following traumatic brain injury were identified (3% of all admissions). The mean age was 44 years and 77% were male. Mortality was 23% in the ICU and 33.5% in the hospital. The median (interquartile range [IQR]) length of stay in the ICU was 3.2 (1.1–8.1) days for survivors and 1.6 (0.7–4.0) days for nonsurvivors. Median (IQR) length of stay in hospital was 24 (10–51) days for survivors and 3 (1–9) days for nonsurvivors. SAPS II, MPM II and the ICNARC model discriminated best between survivors and nonsurvivors and were better calibrated than APACHE II and III in 5393 patients eligible for all models (Table 1).

Table 1

Conclusion

Traumatic brain injury requiring intensive care is associated with a high mortality rate with a short ICU length of stay in nonsurvivors. APACHE II and III have poorer calibration and discrimination than SAPS II, MPM II and the ICNARC model in predicting mortality in these patients.