Interaction Effects Analysis of Platelet Transfusion Data

A statistical method is described for determining which recipient factors influence the efficacy of platelet transfusions in the treatment of thrombocytopenia caused by bone marrow depression. The method, known as interaction effects detection, is useful because it will relate a dependent variable 12‐hour posttransfusion in vivo platelet increment which is numerical to independent variables (recipient factors), which are not necessarily numerical or even rank ordered, in such a manner as to reveal the interactions among the independent variables. The results of the analysis suggest that seven of the 23 independent variables could be key ones in determining the efficacy of a transfusion, as measured by the dependent variable.