Amado Baez, MD, MPH,
Professor and Vice Chair
Medical College of Georgia
Augusta, Georgia, United States
Disclosure information not submitted.
Title: Plasma Lactate and Severity of Illness for Acute Pulmonary Embolism: A Bayesian Probability Model
Introduction: In patients with pulmonary embolism (PE), plasma lactate levels greater than 2mmol/L is associated with higher in-hospital mortality independent of shock and underutilized in risk stratification. Study objective: The objective of the study is to evaluate the added prognostic value of lactate in the assessment of pulmonary embolism by means of a comparative Shock Index and simplified PESI score Bayesian statistical model.
Methods: Shock index (SI) and the simplified pulmonary embolism severity index (sPESI) defined pretest probability. Sensitivity and Specificity for plasma lactate levels were derived from the thrombo-embolism lactate outcome study. Likelihood ratios were calculated and inserted into the nomogram. Bayesian Diagnostic Gains (BDG) were analyzed comparing pre and post-test probability. Bayesian Number Needed to Diagnose (B-NND) were calculated.
Results: 30-day mortality for Shock Index >1 in the low-risk group, was 10.7% versus 24.1% in the high risk group. Whereas sPESI showed a 30-day mortality rates for low risk 1.5% and 10.7% on high-risk patients was. Sensitivity of initial plasma lactate ( >2 ) was 82.4% (95% CI 56.8% to 95.3%) and a specificity of 73.5% (95% CI 71.8% to 74.4%). Integrating lactate values to the Shock Index model demonstrated improved diagnostic gains: SI Low ADG 16.3% (B-NND of 6) with SI high ADG of 25.9% (B-NND of 4). Whereas for sPESI demonstrated a ADG of 3.5% (BNND 29) for low probability and 16.3% (BNND 6) for high pretest probability.
Conclusion: This Bayesian model demonstrated that the combination of shock index and lactate yield superior diagnostic gains than those compared to the sPESI and lactate.