Detecting the determinants of cardiac and hepatic iron overload in patients with thalassemia major using a generalized estimating equations method.

Background and purpose: The diagnosis of iron overload in various organs such as heart and liver in patients with blood-transfusion-dependent thalassemia can greatly improve their quality of life. In this study, we investigated the relationship of T2*-weighted magnetic resonance imaging (MRIT2*) of heart and liver with ferritin levels; also assessed the factors affecting iron overload using generalized estimation equations (GEE). Materials and methods: 53 patients with thalassemia major in Dr. Mohamed Kermanshahi hospital, Kermanshah, Iran, during 2011-12 were enrolled. To investigate the relationship of MRIT2* of heart and liver with serum ferritin levels in addition to linear models, quadratic, exponential and logarithmic data were fitted well. To identify factors influencing iron overload, generalized estimation equations was used. Results: Exponential model was more appropriate to describe the relationship of serum ferritin levels with MRIT2* of heart (R² = 0.170) and quadratic model was more appropriate for liver (R² = 0.277). Generalized estimation equations model showed that ferritin level (P < 0.001), splenectomy status (P < 0.001), desferal (P < 0.001), and osveral (P < 0.001) affected the iron overload. Conclusion: Using generalized estimation equations model for longitudinal studies and studies of correlation between response variables is more reliable and needs smaller sample. [ABSTRACT FROM AUTHOR]