Predictores de sangrado en Hemofilia A severa en una cohorte de pacientes colombianos Thesis

short description

  • Master's thesis

Thesis author

  • Diaz Mosquera, Gina Alejandra
  • Sarmiento Doncel, Samuel

external tutor

  • Pinzon-Rondon, Angela Maria
  • Sierra Hincapie, Gloria Maria
  • Trillos, Carlos Enrique


  • Introducción: In recent decades, hemophilia A management schemes have focused on weight as a dose modifier variable, it is important to identify other associated variables in order to individualize treatment reducing the risk of bleeding. Objetive: To develop an explanatory model to identify the association between anthropometric, behavioral and pharmacokinetic variables with the bleeding phenotype in patients with severe hemophilia A. Methods: We carry out a cross-sectional study, based on the review of 60 clinical records. A logistic regression model was generated to identify the variables that explain the variation in the probability of bleeding. Resultados: The variables that showed association with the bleeding were: age (p 0.003), half-life of FVIII (p 0.027) and physical activity (p 0.000). When adjusting the multiple logistic regression model, it was found that physical activity and age are significant (p 0.001) OR 13.447 ( 95% CI 3.013 – 60.006), and (p 0.019) OR 5.092 ( 95% CI 1.309 – 19.813), respectively. However, when pharmacokinetics were included (p 0,118), e prediction capacity of the model improved with respect to bleeding events. The evaluation of the ROC curve showed an AUC of 0.857, which indicates a good discrimination and a value (p 0.831) for the goodness-of-fit test, which indicates a good calibration. Conclusions: The results suggest that other variables such as age, medication half-life and physical activity can help to find a treatment scheme that correctly discriminates and classifies patients according to health risk, individualizing the behavior and leading to better results in the management of the disease.

publication date

  • September 26, 2018 4:12 PM

Document Id

  • 78564ff9-b75b-45d6-a0ed-73a1569b8a5b