Relationship between illness severity scores in acute kidney injury

BACKGROUND: In the field of critical care nephrology, recent publications have used different illness severity scoring systems, making outcome comparisons difficult. OBJECTIVE: To establish a methodology to translate one illness severity scoring system into another for critically ill patients with acute kidney injury (AKI). DESIGN: Statistical analysis of prospectively obtained data. METHODS: Using data from the Australian and New Zealand Intensive Care Society Adult Patient Database, we obtained Acute Physiology and Chronic Health Evaluation (APACHE) II, APACHE III and Simplified Acute Physiology Score (SAPS II) scores for all patients admitted with AKI. We applied correlation and linear regression analyses as well as cross validation with hold out samples. RESULTS: Between 2001 and 2010, the three illness severity scores were obtained in 636 431 admissions. Of these, 37 203 fulfilled the APACHE score criteria for AKI. The coefficient of determination (R2) between APACHE II and APACHE III scores was 0.66. The overall model was APACHE III = 3.13 ? APACHE II + 7.99 (P < 0.001). Similarly, the R2 between APACHE III and the SAPS II scores was 0.78. The overall model was APACHE III = 1.49 ? SAPS II + 15.5 (P < 0.001). The R2 between APACHE II and SAPS II scores was 0.62. The overall model was APACHE II = 0.35 ? SAPS II + 9.3 (P < 0.001). CONCLUSIONS: Simple, robust translational formulae can be developed to allow clinicians to compare illness severity of patients with AKI when illness severity is expressed with different scoring systems.