Improvement in detecting sepsis using leukocyte cell population data (CPD)
Autor: | Eloísa Urrechaga, Oihane Bóveda, Urko Aguirre |
---|---|
Rok vydání: | 2018 |
Předmět: |
Quality Control
medicine.medical_specialty Wilcoxon signed-rank test Clinical Biochemistry 030204 cardiovascular system & hematology Gastroenterology Sepsis 03 medical and health sciences 0302 clinical medicine Internal medicine Leukocytes Cluster Analysis Humans Medicine Cell Population Data Retrospective Studies Principal Component Analysis Receiver operating characteristic business.industry Biochemistry (medical) Curve analysis General Medicine medicine.disease Blood Cell Count Exact test ROC Curve Area Under Curve 030220 oncology & carcinogenesis business Unsupervised clustering Sysmex xn |
Zdroj: | Clinical Chemistry and Laboratory Medicine (CCLM). 57:918-926 |
ISSN: | 1437-4331 1434-6621 |
DOI: | 10.1515/cclm-2018-0979 |
Popis: | Background The cell population data (CPD) parameters reported by XN analyzers (Sysmex Corporation, Kobe, Japan) reflect the size and internal structure of leukocytes. We explored whether CPD values could contribute to recognize those patients with fever at risk to develop sepsis. A profile of sepsis was developed combining CPD parameters and other markers. Methods We recruited 295 patients at the onset of fever, with infection confirmed by positive cultures. We studied the diagnostic performance of the CPD parameters in the differential diagnosis of sepsis vs. non-systemic bacterial infection using receiver operating characteristic (ROC) curve analysis. Additionally, the K-means unsupervised clustering method was applied. Once the clusters had been defined, the relationship between them and the CPD parameter values was assessed with the non-parametric Wilcoxon test. Lastly, the relationship between the clusters obtained and the categorical variables was examined with the χ2-test (or Fisher’s exact test). Results ROC analysis demonstrated that NE-FSL, NE-WY, NE-WZ and MO-WZ had areas under the curve (AUCs) >0.700 for predicting sepsis. Using the K-means clustering algorithm, 80 patients (66.67%) were assigned to Cluster 1 and the others to Cluster 2. Out of 80 of patients in Cluster 1, 45 (56.25%) presented a PCT value ≥2 ng/mL, whereas almost 80% of Cluster 2 patients had a PCT Conclusions CPD related to monocyte complexity and neutrophil activation were found to be significant, with high values suggesting sepsis. |
Databáze: | OpenAIRE |
Externí odkaz: |