C-reactive protein improves risk prediction in patients with acute coronary syndromes

Abstract

AIMS: Elevated C-reactive protein level is a risk marker in patients with acute coronary syndromes (ACSs), but current risk score systems do not consider this factor. We studied the incremental predictive value of adding C-reactive protein to the Global Registry of Acute Coronary Events (GRACE) risk score. METHODS AND RESULTS: Characteristics, treatments and 30-day mortality were recorded for 1408/1901 consecutive ACS patients. Changes in global model fit, discrimination, calibration, and reclassification were evaluated upon addition of C-reactive protein to the GRACE risk score. High-C-reactive protein patients (C-reactive protein >22 mg/L, 4th quartile of C-reactive protein) were older, had more comorbidities and worse haemodynamic conditions, received less recommended treatment, and had a four-fold higher 30 day mortality. Multivariable analysis demonstrated high-C-reactive protein as an important and independent predictor of mortality. Addition of high-C-reactive protein in the GRACE model modestly improved global fit, discriminatory capacity (c-statistic from 0.795 to 0.823), and calibration. Patients were divided into four groups according to GRACE risk score prediction: <1, 1 to <5, 5 to <10, and >or=10%. The model with high-C-reactive protein allowed adequate reclassification in 12.2%. CONCLUSION: Elevated C-reactive protein level is a modest but independent predictive factor of 30-day mortality in ACS patients, even after adjustment for co-morbidities, haemodynamic conditions, and treatment. Combined with the GRACE risk score, C-reactive protein information improves risk classification.