Internal and external validation of models.
<div><p>Background</p><p>Thromboelostograms (TEG) are indicators that reflect the dynamic changes in blood coagulation objectively. Compared with traditional coagulation tests, TEG are easy to perform; imparting a head start in determining patients’coagulation status. The aim...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | , , , , , , |
| Published: |
2025
|
| Subjects: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | <div><p>Background</p><p>Thromboelostograms (TEG) are indicators that reflect the dynamic changes in blood coagulation objectively. Compared with traditional coagulation tests, TEG are easy to perform; imparting a head start in determining patients’coagulation status. The aim of this study was to explore the role of thromboelastography as an early predictor of disease severity and as a prognostic factor in patients with haemorrhagic fever of renal syndrome (HFRS).</p><p>Methods</p><p>This was a retrospective study in which we collected clinical data from 342 patients with HFRS who were hospitalized from January 2017 to January 2021. The predictive value of laboratory parameters for HFRS criticalization was assessed via receiver operating characteristic (ROC) curves. After that, a model of criticalization was developed using stepwise analysis, subsequently, the model was evaluated and validated internally as well as externally. This study was approved by the Ethics Committee of the Hospital.</p><p>Results</p><p>The study population was categorized into critical and noncritical groups according to their condition during hospitalization, with a median age of 52.00 (39.00–61.50) years for the critical group and 39.00 (16.00–53.25) years for the noncritical group. Both groups had a large proportion of male patients (65.3% and 72.3%, respectively). The median duration of hospitalization was 15.00 (3.50–25.00) days in the critical group and 13.00 (10.00–17.25) days in noncritical group. In the critical group 40.8% of patients died. The incidence of bleeding was greater in critical patients (51.0%) than in noncritical patients (15.1%). The coagulation indices: CI, K, MA, R and angle were significantly different between the two groups (p < 0.05). Critical patients had higher levels of K and R and lower levels of CI, MA and angle. Multivariate analysis revealed that K, ALB, CK, SCR, angle, WBC and LYM were independently associated with disease severity in patients with HFRS. ROC curve analysis revealed that the criticality model equation had an AUC of 0.9058, a sensitivity of 88.9% and a specificity of 80.2%, which were better than that of any single parameter in predicting the patient’s outcome. The internal validation C index was 0.866, the GOF P value was 0.825, and the external validation AUC was 0.762, with a sensitivity of 71.4% and a specificity of 74.2%. The results of the calibration plots indicated that there was some agreement between the model-estimated HFRS critical illness rates and the final observed critical illness rates in the study population. Decision curve analysis indicated that the model had significant clinical utility and comparable net benefit over a range of threshold probabilities.</p><p>Conclusion</p><p>Critical HFRS have significantly abnormal coagulation function and high incidence of bleeding associated with poor outcome. A criticalization prediction model constructed on the basis of thromboelastography indices in early stages of the disease has a good predictive ability and may be helpful in the early identification of critically ill patients for timely clinical intervention and treatment.</p></div> |
|---|