Utility of Blood Cellular Indices in the Risk Stratification of Patients Presenting with Acute Pulmonary Embolism.
Brett SlajusYevgeniy BrailovskyIman DarwishJawed FareedAmir DarkiPublished in: Clinical and applied thrombosis/hemostasis : official journal of the International Academy of Clinical and Applied Thrombosis/Hemostasis (2021)
Pulmonary embolism (PE) clinical manifestations vary widely, and that scope is not fully captured by current all-cause mortality risk models. PE is associated with inflammatory, coagulation, and hemostatic imbalances so blood cellular indices may be prognostically useful. Complete blood count (CBC) data may improve current risk models like the simplified pulmonary embolism severity index (sPESI) for all-cause mortality, offering greater accuracy and analytic ability. Acute PE patients (n = 228) with confirmatory diagnostic imaging were followed for all-cause mortality. Blood cellular indices were assessed for association to all-cause mortality and were supplemented into sPESI using multivariate logistic regression. Multiple blood cellular indices were found to be significantly associated with all-cause mortality in acute PE. sPESI including red cell distribution width, hematocrit and neutrophil-lymphocyte ratio had better predictive ability as compared to sPESI alone (AUC: 0.852 vs 0.754). Blood cellular indices contribute an inflammatory and hemodynamic perspective not currently included in sPESI. CBC with differential is a widely used, low-cost test that can augment current risk stratification tools for all-cause mortality in acute PE patients.
Keyphrases
- pulmonary embolism
- liver failure
- inferior vena cava
- end stage renal disease
- respiratory failure
- ejection fraction
- newly diagnosed
- drug induced
- chronic kidney disease
- low cost
- prognostic factors
- peritoneal dialysis
- aortic dissection
- stem cells
- oxidative stress
- patient reported outcomes
- intensive care unit
- bone marrow
- peripheral blood
- machine learning
- mass spectrometry
- electronic health record
- big data
- deep learning