Zhongguo shuxue zazhi (Jun 2023)

Establishment and evaluation of predictive models of platelet transfusion effect in multi-center tumor patients

  • Gang ZHAO,
  • Fan GUO,
  • Qing LI,
  • Yihua XIE,
  • Jun LI

DOI
https://doi.org/10.13303/j.cjbt.issn.1004-549x.2023.06.003
Journal volume & issue
Vol. 36, no. 6
pp. 471 – 474

Abstract

Read online

Objective To study the platelet transfusion predictive models in tumor patients and evaluate its application effect. Methods A retrospective study was conducted on 944 tumor patients, including 533 males and 411 females who received platelet transfusion in the Affiliated Hospital of Traditional Chinese Medicine of Xinjiang Medical University, the Affiliated Cancer Hospital of Xinjiang Medical University and Kailuan General Hospital from August 2022 to January 2023. Multivariate Logistic regression analysis was used to establish the platelet transfusion predictive models, and Medcalc15.8 software was used to draw the receiver operating curve (ROC) to evaluate the application effect of the prediction model. The actual application effect of models was verified through 162 female clinical cases and 172 male clinical cases. Results The incidence of platelet transfusion refractoriness in tumor patients was 28.9% (273/944), with 33.2% (177/533) in males, significantly higher than that in females [23.4% (96/411)] (P<0.05). Platelet transfusion predictive models: Y1 (female) =-8.546+ (0.581×number of pregnancies) + (0.964×number of inpatient transfusion bags) + number of previous platelet transfusion bags (5-9 bags: 1.259, ≥20 bags: 1.959) + clinical diagnosis (lymphoma: 2.562, leukemia: 3.214); Y2 (male) =-7.600+ (1.150×inpatient transfusion bags) + previous platelet transfusion bags (10-19 bags: 1.015, ≥20 bags: 0.979) + clinical diagnosis (lymphoma: 1.81, leukemia: 3.208, liver cancer: 1.714). Application effect evaluation: The AUC (area under the curve), cut-off point, corresponding sensitivity and specificity of female and male platelet transfusion effect prediction models were 0.868, -0.354, 68.75%, 89.84% and 0.854, -0.942, 81.36%, 77.53%, respectively. Actual application results showed that the sensitivity, specificity, and accuracy of female and male model were 89.47%, 92.74%, 91.98% and 83.72%, 91.47%, 89.53%, respectively. Conclusion There is high incidence of platelet transfusion refractoriness in tumor patients, and the predictive model has good prediction effect on platelet transfusion refractoriness in tumor patients, which can provide reliable basis for accurate platelet transfusion in tumor patients.

Keywords