The objective of this study was to develop a Bayesian clinical decision support mathematical model that can assist in assessing a diagnostic utility integrating the aortic dissection detection risk score (ADD-RS) combined with the diagnostic quality of D-dimer testing.Methods:
Our method uses the Bayes nomogram. Pretest probability scoring for the ADD-RS was obtained using their derived precalculated effects models. Sensitivity, specificity, and positive and negative likelihood ratios (LRs) for D-dimer testing were obtained by meta-analysis. Posttest probability was obtained from Bayesian statistical modeling integrating low, intermediate, and high pretest for the ADD-RS and LRs for D-dimer testing. Relative (RDG) and absolute (AADG) diagnostic gains were calculated.Results:
Pool meta-analysis of D-dimer data demonstrated a sensitivity of 0.97 (95% confidence interval [CI], 0.94-0.99), specificity of 0.56 (95% CI, 0.51-0.60), negative LR of 0.06 (95% CI, 0.03-0.12), and positive LR of 2.43 (95% CI, 1.89-3.12). Bayesian modeling for negative LRs demonstrated posttest probabilities scores of 0.24% for low risk (AADG = 4.06% and RDG = 94.42%), 3.4% for intermediate risk (AADG = 33.1% and RDG = 90.68%), and 7.9% for high risk (AADG = 51.3% and RDG = 86.65%).Conclusion:
The integration of the ADD-RS and D-dimer testing in a decision support scheme suggested rule-out diagnostic value and gains, mostly evidenced in the AADD-RS low and intermediate pretest probability categories. We propose further evaluating the use of this decision support scheme in a prospective model and as a potential triage tool for aortic dissection.