External Validation of Five Scores to Predict Stroke-Associated Pneumonia and the Role of Selected Blood Biomarkers

Background and Purpose: Several clinical scoring systems as well biomarkers have been proposed to predict stroke-associated pneumonia (SAP). We aimed externally competitively validate SAP scores hypothesized that 5 selected would improve performance of these scores. Methods: pooled the data 2 acute stroke studies with identical assessment: STRAWINSKI PREDICT. Biomarkers (ultrasensitive procalcitonin; mid-regional proadrenomedullin; proatrionatriuretic peptide; ultrasensitive copeptin; C-terminal proendothelin) were measured from hospital admission serum samples. A literature search was performed identify prediction then calculated multivariate regression models individual biomarkers. Areas under receiver operating characteristic curves used compare discrimination models. Results: The combined cohort consisted 683 cases, which 573 had available backup samples perform biomarker analysis. Literature identified 9 Our set enabled us calculate area curve 0.543 0.651 for physician determined SAP, 0.574 0.685 probable 0.689 0.811 definite according Pneumonia in Stroke Consensus group criteria. Multivariate improved virtually all predictions, but mostly range an delta 0.05. Conclusions: All patients who develop fair strong capabilities, better when stricter criteria diagnosis applied. provided only limited added predictive value, currently not warranting addition markers Registration: URL: https://www.clinicaltrials.gov . Unique identifier: NCT01264549 NCT01079728.