Effect of blood transfusion on survival among children in a Kenyan hospital.

In Africa, blood transfusions are frequently given to treat severe paediatric anaemia. Because of the risk of HIV transmission, identification of when transfusion will reduce the risk of death for severely anaemic children has become increasingly important. For all children admitted to a Kenyan hospital from October, 1989, to October, 1990, we collected data on clinical presentation, haemoglobin (Hb), receipt of transfusion, and in-hospital survival. Of 2433 admissions, 29% (684) had severe anaemia (Hb less than 5.0 g/dl), and 20% (483) received blood transfusions. Based on laboratory criteria only, children with Hb less than 3.9 g/dl who were transfused had lower mortality than those with Hb less than 3.9 g/dl who were not transfused, but this finding applied only to children transfused on the day of admission (odds ratio [OR] 0.30; 95% Cl 0.14, 0.61) or the day after admission (OR 0.37; 95% Cl 0.14, 1.00). Based on a combination of laboratory and clinical criteria, children with clinical signs of respiratory distress and Hb less than 4.7 g/dl who were transfused had lower morality than those who were not (OR 0.19; 95% Cl 0.09, 0.41). Among children without respiratory distress, there was no association between receipt of transfusion and mortality, irrespective of admission Hb. The frequency of blood transfusion can be reduced and survival enhanced by targeting blood to those children with severe anaemia and clinical signs of respiratory distress, and by using transfusion early in the course of hospitalisation.