Background We sought to assess whether high-sensitivity C-reactive protein (hs-CRP) and pro-B-type natriuretic peptide (NT-proBNP) improve risk prediction when added to an established predictive tool and develop a point-based risk score. Methods Four hundred eleven vascular surgery patients were enrolled. The primary outcome was a composite of death, acute coronary syndromes, pulmonary edema within 30 days of surgery, and postoperative troponin-I elevation. The risk score was developed from a logistic regression model by using an integer-based scoring system. Results The rate of the primary outcome was 18%. Adding both hs-CRP and NT-proBNP to the Revised Cardiac Risk Index led to an increase in C statistic from 0.670 to 0.774. The net reclassification improvement was 0.210 (P = 0.004) and the integrated discrimination improvement was 0.112 (P = 0.0001). In the multivariable regression analysis used to develop the risk score, insulin therapy for diabetes (odds ratio [OR]: 2.8; P = 0.003), open surgery (OR: 1.95; P = 0.027), fibrinogen >377 mg/dL (OR: 2.83; P = 0.001), hs-CRP >3.2 mg/L (OR: 3.85; P <0.0001), and NT-proBNP >221 ng/L (OR: 4.05; P <0.0001) were associated with the primary outcome. There was no statistical evidence of overfit. The C index was 0.82 and the Hosmer-Lemeshow statistic was 1.61 (P = 0.0447). The observed and predicted rates of the primary outcome across quartiles of risk score were highly correlated. Conclusions Hs-CRP and NT-proBNP substantially improve risk prediction when added to an established predictive tool. The biochemical marker-based risk score may be useful for accurately risk-stratifying vascular surgery patients; nonetheless, further validation studies on external datasets are needed before it can be used in clinical practice.
ASJC Scopus subject areas
- Cardiology and Cardiovascular Medicine